This article was downloaded by: [Dicle University]

On: 06 November 2014, At: 14:52

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

HVAC&R Research
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/uhvc20

A finite element model to predict sound

attenuation in lined and parallel-baffle
rectangular ducts
a a
Davide Borelli & Corrado Schenone
a
Department of Mechanical Engineering , Università degli Studi di
Genova , Italy
Published online: 11 Jun 2012.

To cite this article: Davide Borelli & Corrado Schenone (2012) A finite element model to predict
sound attenuation in lined and parallel-baffle rectangular ducts, HVAC&R Research, 18:3, 390-405

To link to this article: http://dx.doi.org/10.1080/10789669.2012.641804

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
 A finite element model to predict sound attenuation in
lined and parallel-baffle rectangular ducts
Davide Borelli and Corrado Schenone∗
Department of Mechanical Engineering, Università degli Studi di Genova, Italy
∗
Corresponding author e-mail: corrado.schenone@unige.it

The present article describes a Finite Element Model (FEM) model for the analysis of sound attenuation
Downloaded by [Dicle University] at 14:52 06 November 2014

in lined and parallel-baffle rectangular ducts. By means of the numerical model it was possible to predict
insertion losses in a frequency range up to 8000 Hz. The model was validated by comparing numerical
results with experimental data obtained in accordance to EN ISO 11691 and EN ISO 7235 standards.
For the lined ducts both lining thickness and physical properties of lining were varied during validation;
the comparison indicates a good accuracy of the numerical model for all different tested conditions.
The results of the validation for parallel-baffle rectangular ducts show that also the behaviour of these
dissipative/reactive silencers, with their internal-reflections and energy dissipation phenomena, can be
predicted quite well by the FEM model on the whole frequency range. Overall comparisons suggest that
FEM modelling can be an accurate and inexpensive way to predict sound attenuation in lined ducts and
parallel-baffle mufflers.

Introduction is made to dissipation of acoustic energy into heat

The problem of reducing noise propagation along
air ducts is an increasingly topical subject in the
HVAC sector, both because of the increased expec-
tation for acoustic comfort and also as a result of
the consequent regulations. This leads to the need
to widen the knowledge about air ducts silencing
and, in particular, to develop appropriate computa-
tional tools.
In general, mufflers should have adequate in-
sertion loss over the frequency range of interest,
minimum flow restriction, moderate size and cost,
high durability, good quality of the exhaust sound,
performance should not deteriorate with time, the
flow generated noise should be sufficiently low, and
should be easy to manufacture and to maintain. In
many air conditioning and ventilation systems resort
by lining the air ducts with acoustically absorptive
materials. Ducts are generally rectangular and are
lined at one, two or, less frequently, at all four walls.
Large fans often have parallel-baffle mufflers made
up of an acoustically absorptive material covered
with thin, perforated metallic or plastic sheets for
protection against flow. Normally fibreglass-type,
highly porous materials are used for the lining.
From the experimental point of view, the Euro-
pean Standards EN ISO 7235:2009 (EN ISO 2009a)
and EN ISO 11691:2009 (EN ISO 2009b) and the
American Standard ASTM E477-06a (ASTM 2006)
set the laboratory measurement procedures for de-
termining the insertion loss, in frequency bands,
of ducted silencers and air-terminal units with
and without airflow. From the theoretical point of
view, in the past, analytical methods (Cremer 1953;

Received April 26, 2011; accepted December 13, 2011

Davide Borelli, PhD, is Postdoctoral Researcher. Corrado Schenone, PhD, is Associate Professor.

390
HVAC&R Research, 18(3):390–405, 2012. Copyright 
C 2012 ASHRAE.

ISSN: 1078-9669 print / 1938-5587 online

DOI: 10.1080/10789669.2012.641804
 HVAC&R RESEARCH
Ingard 1955; Kurze 1969) were used in order to the-
oretically evaluate sound attenuation and to avoid
expensive and time consuming experiments. Ac-
cording to these approaches, the cross dimensions
are small enough to ensure that only the plane waves
propagate; the duct is assumed to be uniform in cross
section and the flow Mach number is assumed to be
small enough and constant across the section of the
duct. Unfortunately, although in many applications
higher modes can be ignored, there are situations
in practice where assumptions made in analytical
theory are violated. Besides, comparisons with ex-
perimental data revealed that such predictions were
not always so accurate as design requires (Bartolini
Downloaded by [Dicle University] at 14:52 06 November 2014
et al. 2002) and the complexity of the analytical
approach tends to limit its application to the com-
mon design practice. On the contrary, a large diffu-
sion of adequate design methods is now necessary
to fulfil the ever rising needs of acoustic comfort
in so many situations involving AC and ventilation
plants.
Computer simulations of sound propagation in-
side air ducts can help the choice of the acous-
tical properties of linings or mufflers and their
geometrical design, giving great advantages over
experiments or analytical methods in terms of
reducing costs and obtaining accurate results in a
faster way. Furthermore, numerical methods permit
to solve acoustic problems for complicated geome-
tries, when analytical methods, including Fourier
series approach, become impractical to use. Appli-
cations range from those of a more traditional kind,
as the control of noise emitted by the induction and
exhaust ducts of boilers or power plants, to sectors
such as air conditioning and ventilation plants.
Various numerical discretization techniques for
acoustic propagation in dissipative ducts were intro-
duced since the Sixties (Gladwell 1965; Zienkiewicz
1971) and, in particular, for acoustic propagation
in mufflers at the half way of Seventies by Young
and Crocker (Young and Crocker 1975; Young and
Crocker 1976). Since then, numerical techniques,
such as finite element (Marburg and Nolte 2008; Ih-
lenburg 1998) and boundary element (Kirkup 1998;
Ciskowski and Brebbia 1991) method for low and
mid frequencies acoustical problems and statisti-
cal method (Lyon and DeJong 1995) for high fre-
quency problems, have encountered a rising suc-
cess. A great number of successful applications in
this field can be counted up to now. In recent years,
some hybrid methods combining Statistical Energy
Analysis (SEA) and FEM (energy finite element
391
method—EFEM) or BEM (high frequency bound-
ary element method—HFBEM) have become quite
popular (Langley and Bremner 1999; Vlahopoulos
and Zhao 1999; Shorter and Langley 2005).
All these computational methods have been ap-
plied specifically to model sound propagation in
lined ducts, including boundary element method
(Juhl 2000), finite element method (Astley and
Cummings 1987; Christiansen and Krenk 1988),
statistical analysis (Zlavog and Eversman 2005) and,
recently, hybrid technique (Buske et al. 2010). In
particular, FEM and BEM have been used more con-
veniently in lined ducts and applied in a very general
way. Moreover, in recent years, some commercial
software, user friendly, has been introduced to the
market, giving many engineers the opportunity to
approach numerical design without a specific skill
in computational acoustics. The strength of these
tools and their limited cost have made it possible for
a growing number of engineers their use in acoustics
and, particularly, in calculating noise propagation in
ducts.
The general applicability of computational meth-
ods in commercial codes, which makes them avail-
able to the industrial users, makes more and more
relevant the issue to assess the adequacy and reliabil-
ity of such an approach. This result can be achieved
by testing the accuracy of the simulations and their
effectiveness in guiding the project. For this pur-
pose, in this research, the sound propagation along
air-ducts lined with sound absorbing material and
parallel-baffle mufflers was modelized by means of
a commercial FEM software. Mineral wool, the ma-
terial used normally for the lining because of its
good sound-absorbing properties, with an acoustic
behaviour typical of porous materials (Bies 1988),
was chosen as sound absorbing material and its
acoustical impedance was calculated by means of
the Delany-Bazley correlation (Delany and Bazley
1970).
A number of methods, at times really sophisti-
cated, have been developed in last decades for the
prediction of the acoustical properties of porous me-
dia (Attenborough 1982; Kidner and Hansen 2008):
analytical (Zwikker and Kosten 1949; Morse and
Bolt 1944; Berryman 1979; Berryman 1980; Biot
1956a; Biot 1956b; Allard and Atalla 2009) and
numerical (Kang and Bolton 1995; Panneton and
Atalla 1997; Tanneau et al. 2006) models were intro-
duced in order to explain the behaviour of such ma-
terials by considering their different structures (Per-
rot et al. 2004; Voronina 1998), shape (Becot et al.
 392 VOLUME 18, NUMBER 3, JUNE 2012
2008; Turo and Umnova 2010) and frames (Rossetti
et al. 2003) and their application in the different
fields. These studies have not been specifically con-
sidered in the present article, with the intent to limit
the interest to a well-consolidated and basic method
like the above mentioned Delany-Bazley model.
This semi-empirical model was subsequently de-
veloped by other Authors (Dunn and Davern 1986;
Qunli 1988; Miki 1990a; Miki 1990b), but in the
present study the original formulation has been
adopted and utilized, because this empirical model
often proved to be adequate for most practical ap-
plications of porous materials for sound absorption.
To validate the FEM model accuracy, the nu-
Downloaded by [Dicle University] at 14:52 06 November 2014
merical results were compared with data measured
by using an experimental set-up based on standards
EN ISO 11691 and EN ISO 7235; the apparatus was
used to measure the insertion loss in straight ducts
with square cross section lined with mineral wool.
The finite element model was finally utilized to pre-
dict the sound attenuation along lined and parallel-
baffle rectangular ducts till to high-frequency, where
FEM becomes critical, also analyzing the effect of
density and thickness of the lining material on the
sound propagation inside lined ducts
Numerical modelling
Governing equations
In a lossless medium, sound waves are governed
by the following equation:


1 ∂2 p 1
+ ∇ · − (∇ p − q) = Q, (1)
ρ0 cs2 ∂t 2 ρ0
where p is the acoustic pressure (whose SI unit is
Pa), ρ 0 (kg/m3) is the density, cs (m/s) is the speed
of sound and q (N/m3) and Q (1/s2) are respectively
the dipole source and the monopole source, both op-
tional. The combination ρ 0 cs 2 is called the adiabatic
bulk modulus, commonly denoted K (Pa).
A special case is a time-harmonic wave, for
which the pressure varies with time as:
p (x, t) = p (x) eiωt , (2)
where ω = 2π f (rad/s) is the angular frequency, with
f (Hz) denoting the frequency. Assuming the same
harmonic time dependence for the source terms,
the wave equation for acoustic waves reduces to an
inhomogeneous Helmholtz equation:


1 ω2
∇ · − (∇ p − q) − p = Q, (3)
ρ0 ρ0 cs2
where p = p(x,ω). This formulation makes possible
to compute frequency response using the paramet-
ric solver to sweep over a frequency range using a
harmonic load, specifying the angular frequency ω
through the frequency f .
For spaces such as rectangular lined ducts, a
plane wave only would propagate (with all higher
modes, even if present, being cut-off) if a frequency
is small enough so that:
cs
f < , (4)
2h
where h is the larger of the two transverse dimen-
sions of the rectangular duct (Munjal 1987). For the
geometries used in this article to validate the model,
this condition is satisfied for frequencies up to about
600 Hz.
The absorption mechanism of mineral wool is
based on the viscous effects in the air contained in-
side the interconnecting pores which separate the
fibres; the solid structure can generally be regarded
as ideally rigid and stationary, and the acoustic ab-
sorption as produced by the viscosity of the air,
which is subjected to a process of alternate com-
pression and rarefaction as a sound wave passes
through it (Zwikker and Kosten 1949; Attenbor-
ough 1993). To model the sound damping by means
of the rock-wool set inside the simmetrical side
pockets of the simulated lined duct, the Delany-
Bazley model for porous media (Delany and Bazley
1970) has been used: in this way, we have that the
speed of sound and the density are frequency de-
pendant complex quantities, denoted by ρ c and cc
respectively:
Z c kc ω
ρc = cc = .
ω kc
The Helmholtz equation becomes:


1 ω2
∇ · − (∇ p − q) − p = Q, (5)
ρc ρc cc2
 HVAC&R RESEARCH 393

and, according to the model of Delany-Bazley, we r the Delany-Bazley damping type has been intro-
have: duced to the subdomains where the lining is.


 
ω ρ0 f −C2 ρ0 f −C4 The matched boundary condition is particu-
kc = 1 + C1 −iC3
cs Rf Rf larly useful for modeling acoustic waves in ducts;
such condition belongs to the class of nonreflect-
(6) ing boundary conditions (NRBCs) (Givoli and Neta

−C6
−C8  2004) and it is a reformulation of the Higdon condi-
ρ0 f ρ0 f
Z c = ρ0 cs 1 + C5 −iC7 . tions for plane waves (Higdon 1994). It allows two
Rf Rf modes with wave numbers k 1 and k 2 to leave the
(7) model domain with minimal reflections, according
to the following Equation 9:
The flow resistivity Rf to be introduced in Equa-

tions 6 and 7 is calculated as a function of apparent 1
− n · − (∇ p − q)
Downloaded by [Dicle University] at 14:52 06 November 2014

density ρ rw and mean fiber diameter drw of mineral ρc

wool (Bies and Hansen 1980):
  
2
i ω
cc
+ k1 k2 p + cTc p
ρr1.53
w K +
Rf = . (8) ρc (k1 + k2 )
dr2w

  
2
In Equations 6 and 7 Ci are coefficients specific i ω
cc
+ k 1 k 2 pi + cTcpi
1
for the used material; in Equation 8 K is an empir- = +n· ∇ pi
ical coefficient equal to 3.18∗ 10−9 in SI units and ρc (k1 + k2 ) ρc
8.75∗ 10−7 in I-P units (Bies 1988). Equation 8 is (9)
a semi-empirical correlation for various classes of
fibrous materials; it derives from Nichols’ equation where n is the wave-direction vector, T refers to the
(Nichols 1947) and it was tested for different sam- Laplace operator in the tangential plane at one given
ple thicknesses and densities. By using Equation point r = (x,y,z) on the boundary, and pi represents
8, flow resistivity may be estimated based upon a an incident pressure field of an optional incoming
knowledge of fibre diameter and bulk density alone. plane wave with wave vector k and amplitude p0
Its validity range largely exceeds the range of den- according to:
sity analyzed in the present article.
pi = p0 e−i(k·r) . (10)
Numerical solution
To numerically solve Equation 2, and to deter- If properly set up, the matched boundary con-
mine the sound attenuation inside the lined duct the dition allows the modelization of both the entrance
time-harmonic analysis model type has been set in a and the ending sections of the duct. In particular, the
commercial FEM software. The elements are of the broadband loudspeaker at the entrance section is in-
Lagrange-quadratic type, and boundary and subdo- tegrated specifying an incoming plane wave with
main conditions have been set as follows: wavenumber k 1 = k 2 = ω/cc , while the 1/3 octave
band analysis is provided by the parametric solver
r the sound-hard boundary condition has been used which solves the problem for the desired frequen-
to model rigid surfaces; cies. The ending section of the duct is simply mod-
r the matched boundary condition has been used elized without inserting in the matched boundary
in the entrance section of the duct to specify the condition the optional incident pressure field previ-
acoustic pressure at the boundary; ously described.
r the matched boundary condition has been used The continuity condition, applied to the inte-
as an appropriate approximation for the ending rior boundaries, represents a situation where the
section of the duct. interior boundary has no effect on the acoustic
r the continuity condition has been used to mod- pressure field, so that we have the continuity of
elize the interface between the lining and the duct; the normal acceleration, represented by means of
 394 VOLUME 18, NUMBER 3, JUNE 2012
Figure 1. Example of brick mesh used for a lining thickness of
50 mm (1.97 in.) (color figure available online).
Equation 11:
Downloaded by [Dicle University] at 14:52 06 November 2014



1 1
n − (∇ p − q) − − (∇ p − q) = 0,
ρ0 1 ρ0 2 the most common softwares and solvers for FEM
(11)
where the subscrips 1 and 2 refer to the two sides of
the considered boundary.
The lining is modelled by applying the Delany-
Bazley damping properties to the elements of the
subdomains representing the volume of the lining
itself.
The mesh used for the simulations is a brick mesh
with cuboid shaped elements (Figure 1), and the
number of Lagrangian elements per wavelenght has
been kept between six to ten in order to obtain the
desired accuracy (Marburg 2002; Marburg 2008;
Ihlenburg and Babuška 1995; Ihlenburg and
Babuška 1997; Perrey-Debain et al. 2004). The
hardware configuration used to run the simulations
was a PC with a 2.80 GHz Intel core i7 860
CPU (64 bit, quad-core with Hyper-Threading
Technology) equipped with 8 GB of RAM. The
operating system chosen was Fatdog64, a Linux
distribution with very low footprint that allows
the FEM solver to have the highest amount of free
memory available for the high RAM consuming
computations (Thompson 2006).
The model was solved at first by using one of the
most commonly employed algorithms for matrices
arising from finite element formulations for the
Helmholtz equation (Von Erstoff et al. 2008); such
solvers are commonly used with preconditioners
that intend to improve the overall conditioning of
the system matrix to obtain a good performance
of the solver itself. The chosen solver has been
the Generalized Minimal RESidual algorithm for
solving nonsymmetric linear systems, GMRES
(Saad and Schultz 1986); the preconditioner used in
the simulations was the Geometric Multigrid one,
which allows it to accelerate the convergence of
the base iterative method by correcting, from time
to time, the solution globally by solving a coarse
problem. In this way, it was possible to obtain
correct results for frequencies less or equal to 2000
Hz; from then on, the software was not able to give
any results giving errors due to lack of RAM, or it
continued endlessly in the iterating steps to solve the
problem without coming to converge. The problem
was completely resolved with the PARallel sparse
DIrect SOlver (PARDISO) (Schenk and Gärtner
2004; Schenk and Gärtner 2006), available with an
updated version of the FEM commercial software,
which made it possible to analyze frequencies
up to 8000 Hz in a much faster way than using
traditional solvers. A frequency upper limit of
8000 Hz is a noticeable achievement, since usually
models allow to solve the numerical problem
up to definitely lower frequencies, relegating to
other specifically developed methods, such as SEA
(Statistical Energy Analysis), the analysis of fre-
quencies above FEM’s limit of predictability (Soize
et al. 1992; Rabbiolo et al. 2004; Łodygowski and
Sumelka 2006; Troclet et al. 2009).
For example, Figure 2 shows the calculated pres-
sure distribution for a lining thickness of 50 mm
(1.97 in.) and a frequency of 400 Hz and sug-
gests that the model correctly simulates the phys-
ical behaviour of a lined duct with losses at the
boundaries, so that the wavefront near the side
walls is bent toward them. It can be noticed that
the wavefronts are planar in the free duct, whereas
they’re deformed in the lined portion of the duct; the
colours are less intense downstream the lining, and
that indicates an abatement of the sound pressure
values.
Experiments
Experimental set-up
An experimental apparatus, compliant with the
EN ISO 11691 and EN ISO 7235 standards, was
used to assess the insertion loss in straight ducts
with a square cross section (300 × 300 mm, 11.81
× 11.81 in.). The apparatus comprised the follow-
ing elements (Figure 3): a loudspeaker, test ducts,
a muffler or a substitution duct, a transition ele-
ment and a reverberation room. Width and height
of the duct were set by considering significant tech-
nical dimensions, which are typical cross section
dimensions used in manufacturing, as well as the
 HVAC&R RESEARCH
Figure 2. Calculated pressure distribution: (a) isometric view; (b) x-y plane; (c) x-y plane detail (color figure available online).
need to limit the overall length of the experimen-
tal apparatus, which is dependent on cross-section
dimensions.
The loudspeaker was fitted inside a duct 1.70
m (5.58 ft) long, made of 20 mm (0.79 in.) thick
Downloaded by [Dicle University] at 14:52 06 November 2014
chipboard panels, and enclosed in a sealed cabinet
filled with mineral wool. The ducts and the tran-
sition element were made up by 1 mm (0.04 in.)
thick galvanised steel sheets. The length of the test
duct was set to be at least half of the wavelength
corresponding to the minimum frequency in the test
range and not less than four times the duct cross-
section diagonal. The substitution duct, which was
designed and built in the same way as the test ducts,
was used in the measurement procedure to define a
reference condition.
The transition element, which enables the ap-
paratus to be connected to the reverberation room,
must have a low reflection coefficient and dissipative
losses close to zero; moreover, it must not produce
any self generated noise that might influence mea-
surements in the test room. The transition element
constructed for the present apparatus is shaped like
a truncated pyramid with an internal angle of 30◦
between the walls, terminating in the reverberation
room.
The reverberating room forms the termination
of the experimental apparatus, in accordance with
Figure 3. Experimental setup: (a) sketch; (b) picture (color figure available online).
395
standard EN ISO 11691. The actual compliance of
the reverberating room with the requirements de-
fined by standard EN ISO 3743-1:2010 (EN ISO
2010) and EN ISO 3743-2:2009 (EN ISO 2009c)
was tested by means of specific experiments on the
whole frequency range that has been analyzed dur-
ing the present study.
In order to limit flanking transmissions as far
as possible, particular steps were taken. With re-
gard to transmission paths through solid media, rub-
ber washers were inserted between the flanges; to
limit transmission through air, the entire test section
was enclosed in sound-proofing panels made of ex-
panded polyurethane with a 1 mm thick lead sheath
inserted in between them.
The muffler consisted in a duct equipped with
two replaceable symmetrical side pockets of vari-
ous thickness (Figure 4) or, alternatively, in a paral-
lel baffled duct. Sound transparent layers were used
for confining the lining materials: this way, it was
possible to vary both the thickness of the lining
and the absorbing material, and therefore to ana-
lyze the effect that thickness and absorbing material
have on insertion loss, that is, on the effectiveness
of the dissipative muffler. The facing was made up
of a perforated metal sheet, with a percent open
area of 50%. Both literature (Bies 1988; Munjal and
Thawani 1997) and experimental tests, performed
 396 VOLUME 18, NUMBER 3, JUNE 2012
Downloaded by [Dicle University] at 14:52 06 November 2014
Figure 4. Substitution duct with replaceable symmetrical side pockets.
by a standing wave tube, indicated that this kind
of facing does not modify the acoustical behav-
ior of the sound absorbing material. The panels of
mineral wool had the same thickness of the side
pockets, so that the porous material was not com-
pressed and stayed in contact with the perforated
sheet.
The sound source consisted of a random noise
generator, an amplifier and a loudspeaker; the power
of the sound produced had to be sufficient to ensure
that the sound pressure level in the reverberation
room is at least 10 dB greater than the background
noise.
Experimental procedure
A pink noise was generated at the closed end of
the channel by means of a broad-band loudspeaker;
then the generated signal was picked up by micro-
phones positioned in three different points inside
the reverberation room, according to EN ISO stan-
dards. By means of a sound analyzer and its data
analysis software, the signal was post-processed to
determine the one third octave band spectrum of the
sound pressure level inside the reverberation room,
which constituted the reference environment for the
measurements.
To evaluate the spectral insertion loss (IL) pro-
vided by the muffler under examination, measure-
ments were taken by alternatively mounting on the
experimental set-up either the muffler itself or the
substitution duct. The difference for each one-third
octave band, in decibels, between the mean sound
pressure level inside the reverberation room before
(SPL1 ) and after (SPL2 ) the insertion of the dissi-
pative muffler along the test duct returned insertion
loss values:
IL = SPL1 −SPL2 . (12)
It must be noted that the standard EN ISO 7235
defines the insertion loss of the test object as “the
reduction in the level of the sound power in the
duct behind the test object due to the insertion of
the test object into the duct in place of a substi-
tution duct,” whereas the transmission loss of an
air terminal unit is defined as “the difference be-
tween the levels of the sound powers incident on and
 HVAC&R RESEARCH
Downloaded by [Dicle University] at 14:52 06 November 2014
Figure 5. FCR silencer model SQ-A-110-600 (color figure avail-
able online).
transmitted through the test object”; nevertheless,
for tests according to standard EN ISO 7235, the
measured insertion loss of a silencer equals its trans-
mission loss, as reported in the paragraph 3.1, note
2 of the above mentioned standard. It is therefore
allowed to model only the lined duct excluding the
reverberating room, since the FEM model evalu-
ates the transmission loss while the experimental
set-up measures the insertion loss, but, for the rea-
son mentioned above, these two quantities can be
considered equal. So the article, in general, refers
to insertion loss, because this is the physical quan-
tity measured in the present experiments; the calcu-
lated value of transmission loss is then assimilated
to IL in the following, because their values actually
coincide.
Figure 6. Comparison between experimental and producer data for silencer model SQ-A-110-600.
397
To verify the behavior of the experimental setup,
insertion loss was determined for a commercial si-
lencer, the SQ-A-110-600 made by the FCR com-
pany (Figure 5); this has a 0.3 × 0.3 m (0.98 × 0.98
ft) cross section, a length of 0.6 m (1.97 ft) and a
single 0.2 m (0.66 ft) wide sound-absorbing baffle
lined with mineral wool. Measured insertion loss
was then compared with the producer’s catalog data
(FCR 2000) in the range 125-8000 Hz, for octave
band: comparison between experimental data and
catalogue values revealed very good agreement, as
shown in Figure 6.
Model validation for lined ducts
To validate the accuracy of the numerical model
the experimental and simulated IL curves were com-
pared along the whole frequency range. Both lining
thickness and physical properties of lining were var-
ied during validation, to determine the accuracy of
the model and its limits. Of course, the flow velocity
was set to zero, as during the experiments.
Effects of lining thickness
Insertion loss was both measured and calculated
for four different lined ducts characterised by a
thickness of the absorbing lining, t, equal to 25, 50,
100, and 150 mm (0.98, 1.97, 3.94, and 5.90 in.).
Insertion loss was determined for the lined duct de-
scribed before when lined with bulk mineral wool
(density 50 kg/m3, 3.125 lb/ft3) and was measured in
the range 250–4000 Hz, for one-third octave band.
 398 VOLUME 18, NUMBER 3, JUNE 2012
Downloaded by [Dicle University] at 14:52 06 November 2014
Figure 7. Comparison between experimental and simulated data for different lining thickness: (a) 25 mm (0.98 in.), (b) 50 mm (1.97
in.), (c) 100 mm (3.94 in.), and (d) 150 mm (5.90 in.).
Figure 7 reports the comparison between experi-
mental and calculated data for each lining thickness
and, in general, indicates a good accordance. The
well-known effect of lining thickness on IL, that IL
curves become more broad band for thicker linings,
is described correctly by the finite element model.
Besides, the numerical model properly predicts the
specific effect of the lining thickness on IL curves:
for lining thicknesses greater than 50 mm (1.97 in.)
a substantial independence from the thickness of the
lining is indicated; on the contrary, for small lining
thickness, attenuation curves show a selective be-
havior, with a peak in the insertion loss close to the
frequency of 1000 Hz and depending on t. For lining
thickness of 25 and 50 mm (0.98 and 1.97 in.) the
finite element model identifies with good precision
the frequency whereby the maximum attenuation is
achieved and predicts IL on the whole range with
a mean absolute error of 1.0 dB, which is a very
small value if compared with the standard accuracy
of predictive models (Marburg and Nolte 2008).
For experimental data presented in this study, the
accuracy of predictive model has been evaluated ac-
cording to two criteria: the fraction of data predicted
to within ±30%, called λ, and the percent absolute
mean error, ε:
N
1 pr edict.value − ex p.value
ε= %.
N i=1
ex p.value
Figure 8 presents comparison between measured
and predicted insertion loss for the various lining
thicknesses; the dotted lines indicate a difference be-
tween predicted and experimental values of ±30%.
The absolute mean error for all four thicknesses re-
sults equal to 1.4 dB, while ε and λ in reference to
the whole data set of the analyzed events are respec-
tively equal to 25% and 67%.
Effects of physical properties of lining
To validate the model other experiments and sim-
ulations were carried out, by changing the density
of the sound absorbent material, which was set to be
equal alternatively to 50, 70, and 100 kg/m3 (3.125,
4.375, and 6.250 lb/ft3). The aim was to test the ca-
pability of the numerical model to correctly predict
the effect of physical properties of lining material
on sound attenuation.
 HVAC&R RESEARCH 399
Downloaded by [Dicle University] at 14:52 06 November 2014

Figure 8. Comparison between measured and predicted insertion loss for different lining thickness (
t = 0.98 in.; • t = 1.97 in.;
 t = 3.94 in.; t = 5.90 in.).

In Figure 9 simulated and experimental IL curves
for lining thickness of 50 mm (1.97 in.) and differ-
ent mineral wool density are reported. Both exper-
imental and calculated diagram show a progressive
shift of the insertion loss peak towards higher fre-
quencies at increasing density, with a simultaneous
lowering of the peak itself. Besides, the model in-
dicates that the mineral wool density influences the
performance of the lined duct about to the frequency
of 1000 Hz, whereas for frequencies below 500 Hz
and over 1250 Hz the insertion loss is slightly af-
fected by it. Both diagrams confirm this effect of
the density of sound absorbing material on insertion
losses. For both simulated and experimental curves,
the sound attenuation presents its peak at 1000 Hz
for a density of 50 kg/m3 (3.125 lb/ft3); at higher
frequencies the sound attenuation appears to rise,
even if by very little, when the density of the min-
eral wool increases. This behaviour, described by
the experimental data, is well interpreted by the nu-
merical model. On the contrary, whereas numerical
model predicts that IL regularly varies at the differ-
ent frequencies in relationship with the lining mate-
rial’s density, experimental data show substantially

Figure 9. Effect of the density (• ρ = 3.125 lb/ft3;  ρ = 4.375 lb/ft3;  ρ = 6.250 lb/ft3) of the mineral wool for a lining thickness
of 50 mm (1.97 in.): a) simulated and b) experimental insertion loss.
 400 VOLUME 18, NUMBER 3, JUNE 2012
Downloaded by [Dicle University] at 14:52 06 November 2014
Figure 10. Comparison between measured and predicted insertion loss for different lining density ( ρ = 3.125 lb/ft3; • ρ = 4.375
lb/ft3;  ρ = 6.250 lb/ft3).
overlapped curves for 70 and 100 kg/m3 (4.375 and
6.250 lb/ft3) and a separated curve for the lower
density.
Figure 10 represents the comparison between
measured and predicted insertion loss for all tested
densities of sound-absorbing material. In this case
the absolute mean error results to be 1.9 dB, ε is
equal to 19% and λ to 82%; as before, the values
of ε and λ are referred to the whole data set of the
analyzed events.
All of the comparative studies point out that the
model is able to effectively predict the sound attenu-
ation in lined ducts as the geometry of the lining and
the physical properties of the sound-absorbing ma-
terial change. This happens for an extremely large
frequency range, up to values of 4000 Hz. Greater
absolute differences were observed at frequencies
close to 1000 Hz, when IL values are higher and
the accuracy of the experimental apparatus is re-
duced because of the combining actions of the flank-
ing transmission and the break out phenomenon,
whose effects became more relevant while IL rises.
This means that the comparison of measured and
calculated value suffers from the uncertainty in
experiments and, obviously, from the higher ab-
solute values corresponding to the same percent
error.
This opens the possibility to get a correct design
and to optimize the lined ducts in function of the
needed behaviour; of course, the FEM model must
be properly set-up and optimized, because of its high
computational cost (Allard and Atalla 2009).
Model validation for
parallel-baffle ducts
Whereas lined ducts are a kind of mufflers which
are purely dissipative, since the cross section of the
duct itself is completely free from any obstacle or
restriction, parallel-baffle ducts operate as dissipa-
tive/reactive silencers, since the front part of the
baffles reflects sound waves back to the source and
sound attenuation is a consequence of internal re-
flections, too.
In order to validate the accuracy of the finite ele-
ment model for this configuration, a simulation was
firstly implemented for the above mentioned FCR
silencer SQ-A-110-600; the frontal and rear metal
sheets of the baffles were simulated as a sound hard
 HVAC&R RESEARCH 401
Downloaded by [Dicle University] at 14:52 06 November 2014

Figure 11. Comparison between experimental and simulated data for silencer model FCR SQ-A-110-600.

boundary, while all the other boundary conditions
were set as described the Numerical Solution sec-
tion. Figure 11 shows the comparison between ex-
perimental and simulated data in octave band from
125 Hz to 8000 Hz. The accuracy of the model
is substantially assessed: the plot of both curves is
similar and the value of the absolute mean error is
2.9 dB. As for lined ducts, the maximum difference
between experimental and calculated value is ob-
served close to the peak of insertion loss, when the
model again overestimates the measured IL.
For a further validation of the model, another
parallel-baffle duct was tested, whose geometry is
shown in Figure 12. In respect of the previous si-
lencer, the free flow area was larger and the sound
reflection less relevant, thus determining different
acoustical conditions and dissipative/reactive bal-
ancing. The parallel-baffle duct had a inner square
cross section of 300 × 300 mm (11.81 × 11.81 in.),
with two baffles inside with frontal cross section of
300 × 50 mm (11.81 × 1.97 in.); the mineral wool
inside the baffles had a density of 20 kg/m3 (1.25
lb/ft3) and to calculate its flow resistivity Rf , a modi-
fied version of the Equation 8, recommended by the
producer of the mineral wool, was introduced in the
model (Rockwool 2005):
R f = 95.35 · ρr1.37
w . (13)
The average deviation between experimental val-
ues and values calculated with the Equation 13 is
equal to 11.7% upon a range of density from 40
to 180 kg/m3 (from 2.5 to 11.25 lb/ft3) (Rockwool
2005). Such precision is less than the reproducibil-
ity of the experimental measurement of the flow
resistivity (between 12.5% and 16.5%) evaluated
during an Inter-Laboratory Test between different
European laboratories (Garai and Pompoli 2003).

Figure 12. Tested silencer: (a) sketch; (b) picture (color figure available online).
 402 VOLUME 18, NUMBER 3, JUNE 2012
Downloaded by [Dicle University] at 14:52 06 November 2014
Figure 13. Comparison between experimental and simulated data for the silencer shown in Figure 12.
In Figure 13 experimental and simulated IL
curves from 125 Hz to 8000 Hz are shown. Also the
behavior of this dissipative/reactive silencer, with its
internal-reflections and energy dissipation phenom-
ena, seems to be very well predicted by the FEM
model on the whole frequency range: the maximum
absolute error is lower than 0.6 dB. This result con-
firms that the model is able to predict sound attenu-
ation for different geometry and number of parallel
baffles, therefore introducing the possibility to ana-
lyze the performance of the silencer during the de-
sign and, moreover, to optimize its geometry and the
physical properties of the sound absorbing material
in relation to the expected insertion loss.
The overall values of ε and λ for the parallel
baffled ducts considered is respectively equal to 15%
and 93%, while the fraction of data predicted to
within ±20% is 64% in the simulated frequency
range.
Conclusions
A finite element model for the analysis of sound
attenuation in lined and parallel-baffle rectangular
ducts was implemented, that is able to operate up
to high frequency and to take into account differ-
ent physical properties of sound absorbing material
and various geometries of baffles. The finite element
model, worked out by means of a commercial soft-
ware, was based on a time-harmonic analysis and
the elements were of the Lagrange-quadratic type.
The mesh used for the simulations was a brick mesh
with cuboid shaped elements, and the number of La-
grangian elements per wavelenght was kept between
six to ten in order to obtain the desired precision.
Equations were solved using the PARDISO solver,
which made it possible to analyze frequencies up to
8000 Hz in a much faster way than using traditional
solvers.
The accuracy of the numerical model was vali-
dated by means of an experimental setup compliant
with standards EN ISO 11691 and EN ISO 7235.
To verify the behavior of the experimental setup,
insertion loss was determined for a commercial si-
lencer: the comparison between measured data and
producer values revealed a very good agreement.
Experimental and calculated insertion losses
were then compared both for the lined ducts and
the parallel-baffle ducts along the whole frequency
range. For the lined ducts both lining thickness and
physical properties of lining were varied during val-
idation. The comparison indicates a good accuracy
of the numerical model for all different tested condi-
tions: the percent absolute mean error for all thick-
nesses was equal to 25% and the fraction of data
predicted to within ±30% was 67%; for the various
densities of sound-absorbing material, ε and λ re-
sulted respectively equal to 19% and 82%. Results
for parallel-baffle rectangular ducts suggest that also
the behaviour of this dissipative/reactive silencers,
with its internal-reflections and energy dissipation
phenomena, can be very well predicted by the fi-
nite element model on the whole frequency range:
the absolute mean error for both mufflers tested was
lower than 1.7 dB.
The overall results indicate that FEM models im-
plemented by means of present commercial soft-
ware are able to predict quite well the behaviour of
 HVAC&R RESEARCH
the rectangular lined ducts and of the parallel-baffle
ducts, for various geometrical and sound absorbing
conditions, along the whole frequency range. Thus,
FEM modelling can be effectively utilized during
the air conditioning and ventilation plants design
to calculate the sound attenuation in such mufflers,
by taking into account both effects of the density
of sound-absorbing material and of the baffles ge-
ometry. Besides, FEM modelling permits to avoid
expensive and time-consuming experiments and to
optimize, with a small effort, the characteristics of
lined and parallel-baffle ducts according to the noise
spectrum and expected sound attenuation.
Downloaded by [Dicle University] at 14:52 06 November 2014
Acknowledgements
This work was developed in the frame of the
collaborative project SILENV—Ships oriented In-
novative soLutions to rEduce Noise & Vibrations,
funded by the E.U. within the Call FP7-SST-2008-
RTD-1 Grant Agreement SCP8-GA-2009-234182.
The authors would like to thank Prof. R. Bartolini
and Dr. P. Bagnerini for their support and assistance.
References
Allard, J.F. and N. Atalla. 2009. Finite element modelling of
poroelastic materials, in Allard, J.F. and N. Atalla (eds.),
Propagation of Sound in Porous Media, Second Edition, pp.
321–23. Chichester: John Wiley & Sons.
Astley, R.J., and A. Cummings. 1987. A finite element scheme
for attenuation in ducts lined with porous material: Com-
parison with experiment. Journal of Sound and Vibration
116(2):239–63.
ASTM. 2006. ASTM E477–06a: Standard Test Method for Mea-
suring Acoustical and Airflow Performance of Duct Liner
Materials and Prefabricated Silencers.
Attenborough, K. 1982. Acoustical characteristics of porous ma-
terials. Physics Reports 82(3):179–227.
Attenborough, K. 1993. Models for the acoustical properties of
air-saturated granular media. Acta Acustica 1:213–26.
Bartolini, R., D. Foppiano, and C. Schenone. 2002. Sound Atten-
uation in Ducts Lined with Granular Material. Proc. Forum
Acusticum 2002, MAT-01-018, Sevilla: EAA–SEA.
Bécot, F., L. Jaouen, and E. Gourdon. 2008. Applications of the
dual porosity theory to irregularly shaped porous materials.
Acta Acustica united with Acustica 94(5):715–24.
Berryman, J.G. 1979. Theory of elastic properties of composite
materials. Applied Physics Letters 35(11):856–58.
Berryman, J.G. 1980. Long wavelength propagation in composite
elastic media 1. Spherical inclusions. Journal of the Acousti-
cal Society of America 68(6):1809–19.
Bies, D.A. 1988. Acoustical properties of porous materials. in Be-
ranek, L.L. (ed.), Noise and Vibration Control., pp. 245–69.
Washington, DC: Institute of Noise Control Engineering.
403
Bies, D.A., and C.H. Hansen 1980. Flow resistance information
for acoustical design. Applied Acoustics 13:357–91.
Biot, M.A. 1956a. Theory of propagation if elastic waves in a
fluid-saturated porous solid. I. Low-frequency range. Journal
of the Acoustical Society of America 28(2):168–78.
Biot, M.A. 1956b. Theory of propagation if elastic waves in
a fluid-saturated porous solid. II. Higher-frequency range.
Journal of the Acoustical Society of America 28(2):179–91.
Buske, C., C. Richter, F. Thiele, C. Yu, and M. Zhuang. 2010.
Validation of a zonal approach computing the sound radiation
from lined ducts. AIAA Journal 48(12):2899–908.
Christiansen, P.S., and S. Krenk. 1988. A recursive finite element
technique for acoustic fields in pipes with absorption. Journal
of Sound and Vibration 122(1):107–18.
Ciskowski, R.D., and C.A. Brebbia. 1991. Boundary elements in
acoustics. Southampton–Boston: Computational Mechanics
Publications and Elsevier Applied Science.
Cremer, L. 1953. Theorie der Luftschall-Dämpfung im Rechteck-
kanal mit schluckender Wand und das sich dabei ergebende
höchste Dämpfungsmaß (in German). Acustica 3:249–63.
Delany, M.E., and E.N. Bazley. 1970. Acoustical properties of
fibrous absorbent materials. Applied Acoustics 3:105–16.
Dunn, I.P., and W.A. Davern. 1986. Calculation of acous-
tic impedance of multi-layer absorbers. Applied Acoustics
19(5):321–34.
EN ISO. 2009a. EN ISO 7235: Acoustics–Measurement proce-
dures for ducted silencers–Insertion loss, flow noise and total
pressure loss.
EN ISO. 2009b. EN ISO 11691: Acoustics–Measurement of in-
sertion loss of ducted silencers without flow–Laboratory sur-
vey method.
EN ISO. 2009c. EN ISO 3743-2: Acoustics–Determination
of sound power levels of noise sources using sound
pressure–Engineering methods for small, movable sources in
reverberant fields–Part 2: Methods for special reverberation
test rooms.
EN ISO. 2010. EN ISO 3743-1: Acoustics–Determination of
sound power levels and sound energy levels of noise sources
using sound pressure–Engineering methods for small mov-
able sources in reverberant fields–Part 1: Comparison method
for a hard-walled test room.
FCR 2000. http://www.fcr.it/E/clima/silenziatori/tecnico/sqx.pdf
(Retrieved on 2011-04-26).
Garai, M., and F. Pompoli. 2003. A European inter-laboratory
test of airflow resistivity measurements. Acta Acustica united
with Acustica 89:471–8.
Givoli, D., and B. Neta. 2004. High-order non-reflecting bound-
ary scheme for time-dependent waves. Journal of Computa-
tional Physics 186:24–46.
Gladwell, G.M.L. 1965. A finite element method for acoustics.
Proc. Fifth Int. Cong. Acoustics, L33. Liege.
Higdon, R. L. 1994. Radiation boundary conditions for dispersive
waves. SIAM Journal on Numerical Analysis 31(1):64–100.
Ihlenburg, F., and I. Babuška. 1995. Finite element solution to
the Helmholtz equation with high wave number Part I: The
h-version of the FEM. Computers & Mathematics with Ap-
plications 30:9–37.
Ihlenburg, F., and I. Babuška. 1997. Finite element solution to the
Helmholtz equation with high wave number Part II: The h-p
 404 VOLUME 18, NUMBER 3, JUNE 2012
version of the FEM. SIAM Journal on Numerical Analysis
34:315–58.
Ihlenburg, F. 1998. Finite element analysis of acoustic scattering.
New York: Springer-Verlag.
Ingard, K.U. 1955. Lined Ducts. In Lukasik, S.J. and Nolle A.W.
(eds.), Handbook of Acoustic Noise Control, Volume I Physi-
cal Acoustics—Supplement 1, pp. 217–40. Springfield (OH):
Wright Air Development Center.
Juhl, P. 2000. Non-axisymmetric acoustic propagation in and ra-
diation from lined ducts in a subsonic uniform mean flow: An
axisymmetric boundary element formulation. Acta Acustica
United with Acustica 86(5):860–9.
Kang, Y.J., and J. S. Bolton. 1995. Finite element modeling of
isotropic elastic porous materials coupled with acoustical
finite elements. Journal of the Acoustical Society of America
98(1):635–43.
Downloaded by [Dicle University] at 14:52 06 November 2014
Kidner, M.R.F., and C.H. Hansen. 2008. A comparison and
review of theories of the acoustics of porous materials.
International Journal of Acoustics and Vibrations 13(3):
112–9.
Kirkup, S.M. 1998. The boundary element method in acoustics.
Heptonstall: Integrated Sound Software.
Kurze, U. 1969. Schallausbreitung im kanal mit periodischer
Wandstruktur (in German). Acustica 21:74–85.
Langley, R.S., and P. Bremner. 1999. A hybrid method
for the vibration analysis of complex structural–acoustic
systems. Journal of the Acoustical Society of America
105(3):1657–71.
Łodygowski, T., and W. Sumelka. 2006. Limitations in ap-
plication of finite element method in acoustic numerical
simulation. Journal of Theoretical and Applied Mechanics
44:849–65.
Lyon, R.H., and R.G. DeJong. 1995. Theory and Application
of Statistical Energy Analysis, 2nd Edition. Oxford: Butter-
worth Heinemann.
Marburg, S. 2002. Six elements per wavelength. Is that enough?
Journal of Computational Acoustics 10:25–51.
Marburg, S. 2008a. Discretization requirements: How many ele-
ments per wavelength are necessary?, in Marburg, S. and B.
Nolte (eds.), Computational Acoustics of Noise Propagation
in Fluids, pp. 309–32. Berlin: Springer-Verlag.
Marburg, S., and B. Nolte. 2008b. Computational Acoustics of
Noise Propagation in Fluids, Marburg, S. and B. Nolte (eds.).
Berlin: Springer-Verlag.
Miki, Y. 1990a. Acoustical properties of porous
materials—Modifications of Delany-Bazley models.
Journal of the Acoustical Society of Japan English Edition
11(1):19–24.
Miki, Y. 1990b. Acoustical properties of porous materials—
Generalizations of empirical models. Journal of the Acousti-
cal Society of Japan English Edition 11(1):25–8.
Morse, P.M., and R.H. Bolt. 1944. Sound waves in rooms. Re-
views of Modern Physics 16(2):69–150.
Munjal, M.L. 1987. Acoustics of ducts and mufflers, pp. 6–9.
New York: John Wiley & Sons.
Munjal, M.L., and P.T. Thawani. 1997. Effect of protective layer
on the performance of absorptive ducts. Noise Control Engi-
neering Journal 45(1):14–8.
Nichols, R.H. 1947. Flow-resistance characteristics of fibrous
acoustical materials. Journal of the Acoustical Society of
America 19(5):866–71.
Panneton, R., and N. Atalla. 1997. An efficient finite element
scheme for solving the three dimensional poro-elasticity
problem in acoustics. Journal of the Acoustical Society of
America 101(6):3287–97.
Perrey-Debain, E., O. Lagrouche, P. Bettess, and J. Trevelyan.
2004. Planewave basis finite elements and boundary elements
for three-dimensional wave scattering. Philosophical Trans-
actions of the Royal Society 362:561–77.
Perrot, C., R. Panneton, and X. Olny. 2004. From microstruc-
ture to acoustic behaviour of porous materials. Canadian
Acoustics–Acoustique Canadienne 32(3):18–9.
Qunli, W. 1988. Empirical relations between acoustical properties
and flow resistivity of porous plastic open-cell foam. Applied
Acoustics 25(3):141–8.
Rabbiolo, G., R.J. Bernhard, and F.A. Milner. 2004. Definition of
a high-frequency threshold for plates and acoustical spaces.
Journal of Sound and Vibration 277:647–67.
Rockwool 2005. http://www.rockwool.it/files/RW-IT/file/
Download/Guida-al-SW.pdf (in Italian, Retrieved on
2011-04-26).
Rossetti, S., P. Gardonio, and M.J. Brennan. 2003. Mod-
elling of the acoustic behaviour of rigid frame porous ma-
terials. Proceedings of the Tenth International Congress
on Sound and Vibration, Stockholm, Sweden: pp. 4665–
72.
Saad, Y., and M. H. Schultz. 1986. GMRES: A generalized min-
imal residual algorithm for solving nonsymmetric linear sys-
tems. SIAM Journal on Scientific and Statistical Computing
7:856–69.
Schenk, O., and K. Gärtner. 2004. Solving unsymmetric sparse
systems of linear equations with PARDISO. Journal of Future
Generation Computer Systems 20:475–87.
Schenk, O., and K. Gärtner. 2006. On fast factorization pivot-
ing methods for symmetric indefinite systems. Elec. Trans.
Numer. Anal. 23:158–79.
Shorter, P.J., and R.S. Langley. 2005. Vibro–acoustic analy-
sis of complex systems. Journal of Sound and Vibration
288(3):669–99.
Soize, C., A. Desanti, and J.M. David. 1992. Numerical methods
in elastoacoustics for low and medium frequency ranges. La
Recherche Aérospatiale English Edition 5:25–44.
Tanneau, O., P. Lamary, and Y. Chevalier. 2006. A boundary
element method for porous media. Journal of the Acoustical
Society of America 120(3):1239–51.
Thompson, L.L. 2006. A review of finite-element methods for
time-harmonic acoustics. Journal of the Acoustical Society
of America 119:1315–30.
Troclet, B., B. Hiverniau, M.N. Ichchou, L. Jezequel, K. Kay-
vantash, T. Bekkour, J.B. Mouillet, and A. Gallet. 2009.
FEM/SEA Hybrid Method for Predicting Mid and High Fre-
quency Structure-Borne Transmission. The Open Acoustics
Journal 2:45–60.
Turo, D., and O. Umnova. 2010. Time domain modelling of sound
propagation in porous media and the role of shape factors.
Acta Acustica united with Acustica 96(2):225–38.
 HVAC&R RESEARCH
Vlahopoulos, N., and X. Zhao. 1999. Basic development of hy-
brid finite element method for midfrequency structural vibra-
tions. AIAA Journal 37(11):1495–505.
Voronina, N. 1998. An empirical model for elastic porous mate-
rials. Applied Acoustics 55(1):67–83.
Von Erstoff, O., S. Petersen, and D. Dreyer. 2008. Efficient In-
finite Elements based on Jacobi Polynomials, in Marburg,
S. and B. Nolte (eds.), Computational Acoustics of Noise
Propagation in Fluids, pp. 231–50. Berlin: Springer-Verlag.
Young, C-I J., and M.J. Crocker. 1975. Prediction on transmission
loss in mufflers by the finite elements method. Journal of the
Acoustical Society of America 37(1): 144–48.
Downloaded by [Dicle University] at 14:52 06 November 2014
405
Young, C-I.J. and M.J. Crocker. 1976. Acoustic analysis,
testing and design of flow-reversing muffler chambers.
Journal of the Acoustical Society of America 60(5):
1111–8.
Zienkiewicz, O.C. 1971. The Finite Element Methods in Engi-
neering Science. London: McGraw-Hill.
Zlavog, G., and W. Eversman. 2005. Statistical distributions for
acoustic power of broad band noise in lined ducts. Collec-
tion of Technical Papers—11th AIAA/CEAS Aeroacoustics
Conference, Vol. 2, pp. 1258–74.
Zwikker, C., and C.W. Kosten. 1949. Sound Absorbing Materials.
New York: Elsevier.