This article was downloaded by: [Memorial University of Newfoundland]

On: 18 July 2014, At: 01:55

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

Journal of Building Performance Simulation

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tbps20

Simulation of acoustical parameters in rectangular

churches
a
Umberto Berardi
a
Department of Civil & Environmental Engineering, Worcester Polytechnic Institute, 100
Institute Road, Worcester, MA 01609-2280, USA
Published online: 26 Jan 2013.

To cite this article: Umberto Berardi (2014) Simulation of acoustical parameters in rectangular churches, Journal of Building
Performance Simulation, 7:1, 1-16, DOI: 10.1080/19401493.2012.757367

To link to this article: http://dx.doi.org/10.1080/19401493.2012.757367

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
 Journal of Building Performance Simulation, 2014
Vol. 7, No. 1, 1–16, http://dx.doi.org/10.1080/19401493.2012.757367

Simulation of acoustical parameters in rectangular churches

Umberto Berardi*
Department of Civil & Environmental Engineering, Worcester Polytechnic Institute, 100 Institute Road, Worcester,
MA 01609-2280, USA
(Received 11 September 2012; final version received 6 December 2012)

Computer simulations to predict acoustical parameters have been attracting renewed interest in recent years. This
paper reports the results of computer simulations of the acoustics of box-shaped churches. Twenty-five buildings
have been considered by varying the dimensional ratios between length, width and height. The study focuses on the
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

values assumed by three acoustical parameters, the clarity, the early lateral energy fraction and the centre time, in
each of the simulated buildings. The parameters of analysis allowed to evaluate the dependence between the shape of
the building and its acoustics. Moreover, in each church, simulations were realized by moving the sound source in
five different positions in order to investigate how the acoustics changes. Considering the trade-off among
dimensional ratios of the room, volume and source position, formulae which predict acoustical parameters in box-
shaped churches are presented. A first validation of these formulae has been made using measured acoustical
parameters in five Italian churches.
Keywords: room acoustic simulation; church acoustics; ray tracing algorithm; rectangular buildings; sound
absorption; scattering

1. Introduction to church acoustics depending on their shape is useful for developing

Architectural acoustics is receiving an increasing preliminary indications and general formulae regard-
attention over the last several years. New technologies ing the influence of architectural design on church
for acoustic simulations and measurements have been acoustics.
developed, and it is now possible to predict and The study of room acoustics requires the under-
measure the acoustics of a room with good accuracy. standing of how the sound energy diffuses and varies in
This paper focuses on the architectural acoustics of a a room. The shape, volume, the acoustical properties
particular type of buildings, namely churches. of the material (sound absorption and scattering) of
The church acoustics represents an interesting each surface and their location in the enclosure play a
research field, as demonstrated by the increasing leading role in acoustics. The lack of knowledge about
number of studies of this topic. Many surveys these properties has often been pointed as a limit to
regarding church acoustics have contributed to create determine the acoustics of a building (Barron 1993).
a consistent body of knowledge in Portugal (Carvalho However, it should be noted that the lack of simple
1994), Spain (Sendra et al. 1999), Switzerland (Desar- formulae for acoustical parameters is another
naulds 2002), Poland (Engel and Kosala 2005), Italy challenge.
(Magrini and Ricciardi 2003, Cirillo and Martellotta Reverberation is the most considered parameter in
2006) and Japan (Soeta et al. 2012). Previous surveys architectural acoustics. It can be described as the
have shown that the acoustics of churches is largely persistence of sound after a source has stopped
different from that of other large spaces, such as (Barron 1993). The objective parameter for measuring
concert halls or theatres. While surveys in existing reverberation is the reverberation time (T), which in
buildings are growing, there is an increasing use of live rooms, where diffuse sound conditions prevail, can
software simulations to understand the acoustics of be estimated by the Sabine’s formula:
churches (Weitze et al. 2002, de Sant’Ana and Zannin
2011, Segura et al. 2011). T ¼ 0:161  V=A ð1Þ
Unfortunately, simulations often are time consum-
ing and difficult to make. For this reason, the where V is the total volume of the room and A is the
investigation of how the church acoustics varies total sound absorption in the room. The total sound

*Email: [email protected]

Ó 2014 International Building Performance Simulation Association (IBPSA)

 2 U. Berardi

absorption (A) is frequency dependent and can be sound, especially affecting the early reflections where
obtained by summing the product of each surface area the energy value is greater.
(Si) with the absorption coefficient (ai) of any The Barron and Lee’s theory has hence been
room material at each frequency (A ¼ S1 6 a1 þ S2 6 reinterpreted to be applied to churches (Galindo et al.
a2 þ . . .). In addition, total absorption should include 2005, 2009, Cirillo and Martellotta 2005, Martellotta
the air absorption which is proportional to the room 2009). According to these reinterpretations, the time at
volume by means of a coefficient (4m) which varies as a which the decay began to be linear is later, the farther the
function of air temperature and relative humidity. measurement position is from the source. Furthermore,
However, this coefficient generally has a significant at points near the source, the early reflections bring more
effect at higher frequencies only (Barron 1993). energy than the ideal classical reverberant field while
In diffuse rooms, the reverberation time has the conversely, when the distance from the source grew, the
advantages of being steady throughout the space and early reflections become weaker.
predictable with the simple formula in Equation (1). In Berardi et al. (2009) have reviewed the new
fact, the Sabine’s formula has been validated by more modified models of the ‘revised theory’. Moreover,
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

than one century of use. Consequently, looking at the they have proposed a way to calculate the sound
acoustics of a church, it is common to refer to the strength and other energy sound-related parameters at
reverberation time to get preliminary information each point of a church with the knowledge of few
about its acoustics. However, as the interest in architectural parameters. On the basis of extensive
acoustics of churches increases, more importance is surveys, Berardi et al. (2009) have proposed to
recognized to characterize these buildings beyond the categorize the churches according to the shape of the
reverberation time. altar zone and the main materials inside the church.
A number of other acoustic parameters have been These parameters, together with the total surface and
defined in order to better match subjective perception volume of the church, allow us to calculate the sound
of the sound field, even though the prediction of such energy decay in the building, and from this decay to
acoustic parameters depends on many factors, such as determine other acoustical parameters, such as the
the relative position of sources and receivers. clarity and the centre time. The formulae proposed by
According to the classical theory of diffuse sound Berardi et al. (2009) offer a way to know how energy-
propagation in rooms with uniformly distributed related acoustical parameters vary through a church.
absorption, the sound pressure level is the sum of a Unfortunately, these formulae are not able to give
direct component and a diffuse one. Barron and Lee enough information about the acoustics of a church
(1998) proposed a ‘revised theory’ assuming that the from its preliminary design because they require the
reflected sound cannot arrive earlier than direct sound. knowledge of the acoustical and geometrical charac-
They proposed a model based on the following four teristics of the altar and its surrounding surfaces.
assumptions. (i) The direct sound is followed by linear This paper investigates how acoustical parameters
level decay at a rate corresponding to the reverberation vary in box-shaped enclosures and tries to define
time. (ii) The instantaneous level of the late decaying simple formulae for preliminary estimation of energy-
sound is uniform throughout the space. (iii) The direct related acoustical parameters. The paper presents the
sound reaches a point at a distance r from the source results of simulations of the acoustics of 25 churches.
after a time proportional to r. In this way, the In particular, the study focuses on the values assumed
integrated energy decreases when the source-receiver by three acoustical parameters, the clarity (C80), the
distance increases, while the early/late reflected energy early lateral energy fraction (JLF) and the centre time
ratio remains constant (iv). (Ts), in each simulated church.
Unfortunately, several studies show that churches The C80 is defined as the ratio, expressed in
can hardly be included among the proportionate decibels, of the early energy (0–80 ms) divided by the
spaces, and the Barron and Lee’s theory is not accurate late reverberant energy (80 ms to infinite). The JLF is
in these rooms (Sendra et al. 1999). Measurements of defined as the ratio between the fractions of energy
both strength and clarity carried out in Mudejar- arriving from lateral directions within the first 80 ms
Gothic churches, in mosques and in Italian churches (pL) over the total energy from 0 to infinity. The Ts is
show that reflected sound level is below that predicted defined as the time of the centre of gravity of the
by previous theories. Possible explanations of this squared impulse response of the sound energy (ISO
difference may be found in the ‘disproportionate’ 3382-1 2009). The formulae for the three parameters
nature of these kinds of buildings, in the non-uniform considered in this paper are:
distribution of sound absorbing materials and in their R 80 2
p ðtÞdt
architectural elements such as side aisles, chapels, C80 ¼ 10 log R01 2 ðdBÞ: ð2Þ
vaults and domes. These elements scatter or hinder the 80 p ðtÞdt
 Journal of Building Performance Simulation 3
R 80 2
p ðtÞdt could be the object of the study. Considering the shape
JLF ¼ 10 log R51 L2  100ð%Þ ð3Þ
0 p ðtÞdt
of existing churches, it is possible to recognize few
common typologies (Strappa 2005). Initially, Christian
R1 2
tp ðtÞdt churches were often built according to a basilica
Ts ¼ 10 log R01 2 ðmsÞ: ð4Þ typology, with an axial plan and pews in rows oriented
0 p ðtÞdt
towards the altar. Churches with more complex
typologies were then adopted through the centuries.
The C80 measures the balance between the useful and More complex shapes seemed to be justified by the
detrimental sound for the listening perception. Mar- symbolism and the new attention for decorations in
shall (1994) suggested values above 76 dB for listen- churches. In fact, a frequently adopted form became
ing organ music. These values, which are much lower the cross shape (given by the main nave and the
than those of concert halls, are common in churches. transept), with three or five lateral naves and chapels.
However, considering the speech transmissibility which More recently, churches have been designed with a
should also be guaranteed in churches and the several square plan with the worshipers organized around the
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

instruments which can performed in these spaces, altar. Another recent typology is the transversal plan,
values of C80 above 73 dB should be preferred which has a larger width than the length of the church.
(Marshall 1994, Berardi 2012). These new typologies are often considered particularly
The JLF is correlated with the perception of coherent with the principles of the Second Vatican
spaciousness of a sound source and with the apparent Council (Strappa 2005). In fact, this has suggested a
source width. The JLF is generally expressed in liturgy which can be better followed by churchgoers
percentage, and its preferred values in churches are with the priest who has to face the congregation.
above 15–20% (Cirillo and Martellotta 2006, Martel- Finally, an unobstructed and smaller distance between
lotta 2008). the priest and churchgoers is largely considered in new
The Ts is related to the balance between early churches (Strappa 2005).
sound and delayed sound respect to the Ts value. It is Having considered the historical churches, it was
expressed in milliseconds, and it can assume values up decided to simulate box-shaped churches only. In
to several hundreds of milliseconds (Berardi et al. particular, different ratios between the length (the axis
2009). In fact, churches generally are reverberant that comes from the entrance to the altar) and the
spaces in which late reflections reach listeners with width were considered. In this way, different shapes
long delay. This late sound contributes to the high were selected from stretched longitudinal churches
reverberation of the room and may improve the (length to width ratio 2:1), longitudinal (*1.3:1),
perception of the music. However, it is generally square (1:1), transversal (*1:1.3) and stretched trans-
detrimental for listening of speeches or the spoken versal (1:2) shapes. Figure 1 shows the shapes that were
word. Preferable values of Ts should hence balance selected for this study.
these opposite requirements: values below 260 ms are The dimensions of a church are generally related to
generally suggested (Fasold and Veres 2003, Berardi the community of churchgoers who use it (CEI 2007).
2012). In particular, the total volume of a church is
The selected parameters of analysis are particularly particularly important for its acoustics, as shown by
dependent from the shape of the church and show Equation (1).
significant variations inside a room. This study will In historical churches, the length and width were
evaluate how acoustics varies inside box-shaped often related to height, because proportional and
buildings only. symbolic ratios between the dimensions were common
The paper is organized according to five sections. features in the past (Strappa 2005). Moreover,
Section 2 outlines the typologies of churches which construction technologies often related length, width
were selected for the acoustical simulations. Section 3 and height. Following the same approach, the pro-
describes the simulations. Section 4 presents and portionality between the dimensions was followed for
discusses the results of the simulations and, finally, establishing the dimensions of the churches to simulate
Section 5 contains concluding remarks. for the present study.
In order to map the variation of the acoustical
parameters in churches with the same shape, five
different volumes were selected: 2500, 5000, 10,000,
2. Definition of churches to simulate
20,000 and 50,000 m3. Table 1 shows the dimensions of
2.1. Selection of the plans each simulated church.
An historical investigation of Christian churches was After having established the shape of the churches,
performed to define the typologies of churches that the internal organization of the space was defined. For
 4 U. Berardi
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

Figure 1. Plans of the churches, with positions of the sources (letters) and receivers (numbers), object of simulations.

this, a different organization for each typology of . the stretched transversal typology had an orga-
church was selected according to the following: nization with a radial pattern and pews arranged
on semi-circles around the altar.
. the stretched longitudinal typology was orga-
nized in a longitudinal way with pews in two lines After having defined the internal organization and
oriented towards the altar which was at the end the positions of the pews and the altar, the locations of
of the longitudinal axis; the sound sources were selected. In churches, two kinds
. the longitudinal typology was organized with of sounds have to be taken into consideration mainly:
pews arranged in three columns oriented towards one is the priest’s voice, and the other is the organ
the end of the church; music (or of any other instrument). This paper only
. the squared typology had a central organization considers the organ music because the priest’s voice is
with pews organized on the four sides around the commonly amplified and diffused into a church
altar; through audio systems. The variability of modern
. the transversal typology had a central organiza- audio systems makes impossible any generalization
tion with pews arranged in three sides of the about the effects of the position of the source. Meyer
church around the altar which was near to the (2003) identified and categorized several positions of
end wall; the organ in historical buildings. Following previous
 Journal of Building Performance Simulation 5

study, five different positions of the source were organ positioned over the central nave, in the
selected for each simulation. These positions were: matroneum or suspended on a lateral wall;
. position C, between the altar and the worshipers.
. position A, above the entrance of the church. This location is considered the most suitable
This location was common before the Second according to the Second Vatican Council as it
Vatican Council, when often the organ was on a places the music source between the celebrant
platform above the main door (Strappa 2005); and the worshipers (CEI 2007);
. position B, in the centre of the room, a location . position D, at one side of the altar in an
which historically corresponded to have the asymmetrical position, similar to position C;
. position E, in line with the longitudinal axis of
the church and suspended behind the altar, as it
can be seen in many choirs located in the apse of
Table 1. Dimensions of the churches object of simulations. churches.
Length Width Height Volume
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

Church typology (m) (m) (m) (m3) The height of the source in each position was
different: it was 2 m from the floor in the position D, at
Stretched 25.0 12.5 8.0 2,500 50% of the height of the church in the positions B, C
longitudinal 33.5 16.8 9.0 5,050
41.0 20.5 12.0 10,086 and E, and at 75% of the total height of the church in
50.0 25.0 16.0 20,000 the position A. Figure 2 shows existing churches with
64.5 32.3 24.0 49,923 the organ in the five positions selected for the
Longitudinal 20.0 15.5 8.0 2,480 simulations.
27.0 20.5 9.0 4,982
33.5 25.0 12.0 10,050 Finally, the simulations required to establish the
40.5 31.0 16.0 20,088 receiver points. Ten receivers were positioned through
52.0 40.0 24.0 49,920 the churches and were placed according to the guide-
Square 17.8 17.8 8.0 2,521
23.5 23.5 9.0 4,970
lines for acoustical surveys in churches (Martellotta
29.0 29.0 12.0 10,092 et al. 2009). In particular, eight positions were in the
36.0 36.0 16.0 20,736 area of the pews, whereas two positions were behind
45.8 45.8 24.0 50,234 the altar and in the back of the church (Figure 1), to
Transversal 15.5 20.0 8.0 2,480
20.5 27.0 9.0 4,982 consider the position of the priest and also the
25.0 33.5 12.0 10,050 churchgoers standing in the back of the church.
31.0 40.5 16.0 20,088
40.3 52.5 24.0 50,715
Stretched 12.5 25.0 8.0 2,500 2.2. Selection of materials
transversal 16.8 33.5 9.0 5,050
20.5 41.0 12.0 10,086 After having defined the typologies of the churches, it
25.0 50.0 16.0 20,000 was necessary to select the materials of each surface. In
32.3 64.5 24.0 49,923 this regard, common materials in existing churches

Figure 2. Existing churches with the organ in the five positions selected for the simulations: Jubilee Church Dives in
Misericordia (Rome), St. Michael’s Church (Luneburg, Germany), Basilica of Santa Sabina all’Aventino (Rome), Padre Pio
Pilgrimage Church (San Giovanni Rotondo, Italy) and Basilica of San Lorenzo (Florence) (source: Cirillo and Martellotta,
2006).
 6 U. Berardi

were considered. Obviously, the selection of the . brick absorbs up to 15%, depending on the mix,
materials is particularly important for an acoustic surface and thickness of the mortar (Barron
simulation, as material properties (in particular, sound 2003);
absorption and scattering) have a large influence over . gypsum absorbs up to 5%, with more absorption
room acoustics (Vorländer 2008). For example, soft- at higher frequencies (Barron 2003);
ware simulations of the church of Hagia Sophia in . plaster, often used in churches for ornamenta-
Istanbul have found that the presence of carpets and tion, has an absorption up to 8%;
curtains which sometimes covered the marble of the . suspended ceilings (false ceiling and wood
church, determined a difference of more than 2 s of the panels) dissipate acoustic energy through vibra-
reverberation time (Weitze et al. 2002). This finding tions at low frequencies and can have an
underlines the impact that materials have on the absorption up to 30%, which only occurs around
acoustics of a church. the resonance frequency of the panel;
The materials of churches generally have low sound . pews absorb from 10 to 20% of the incident
absorption coefficients (Sendra et al. 1999, Galindo sound in unoccupied conditions. Differently,
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

et al. 2009, Segura et al. 2011). High reflective when occupied by churchgoers, pews absorb
materials, such as marble and stone, are common in from 20 to 30% at low frequencies and up to
historical churches and still cover large surfaces in new 80% at high frequencies (Martellotta and Cirillo
buildings (Desarnaulds 2002). However, the presence 2009);
of curtains, carpets, absorbent pews, gypsum and . curtains and carpets have a high absorption
wooden panels determines a high absorption (Berardi which can reach 70% above 1 kHz.
2011).
An important aspect in simulations is the depen- Sources of data for absorption coefficients of
dence of the acoustical properties of a material from materials commonly used in churches are the books
the frequency (Vorländer 2008). Acoustical properties by Meyer (2003) and Cirillo and Martellotta (2006).
are generally considered at least in the frequency bands For example, the plaster ceiling of the church of the
from 100 Hz to 3150 kHz (ISO 3382-1 2009). Common Gran Madre di Dio in Turin has absorption between
materials in churches and their acoustic absorptions 0.20 at 125 Hz and 0.23 at 4 kHz. Moreover,
are: decorated surfaces of marble or plaster like those
found in the church of San Lorenzo in Turin, or in the
. marble absorbs only 2% of incident sound church of San Luca e Martin in Rome, have a mid-
energy, almost independently from the frequency frequency absorption ranging between 0.08 and 0.10
(Meyer 2003); according to the richness of the decorative pattern
. concrete, which is common in modern buildings, (Cirillo and Martellotta 2006). These literature data
absorbs less than 4% (Barron 1993); were considered in selecting the absorption coefficients
. stone behaves differently according to their in the simulations. Finally, Figure 3 shows few
surface finishes, but it generally absorbs between materials commonly used in churches with their
4 and 8% (Meyer 2003); absorption coefficient laws.

Figure 3. Sound absorption of some materials commonly used in churches: marbles, stones, pews (empty in solid and occupied
dashed), glass windows and bricks (data taken from Barron 1993, Vorländer 2008, Martellotta 2009, Martellotta and Cirillo
2009, Berardi 2011).
 Journal of Building Performance Simulation 7

3. Acoustic simulations reflections (although considering angular distribution

This section presents and discusses the input data that in CATT-Acoustics and directional secondary sources
were used to perform the simulations of the acoustics in ODEON). Finally, in this study, CATT-Acoustics
of the considered churches. Five typologies of churches was preferred for its ability to include frequency-
and five volumes were considered, and for each of the varying diffusion coefficients.
25 churches, five different positions of the source were The simulations were performed in the octave
evaluated, obtaining a total of 125 simulations. bands ranging from 125 Hz to 4 kHz, with tempera-
The software CATT-Acoustics v.8 was used for the ture set at 208C and relative humidity at 50%. The
simulation (Dalenbäck 2008). This software is a room number of rays for coverage mapping in each octave of
acoustics prediction program based on a mixed analysis was 5000, the time of ray truncation was 2 s
algorithm. First, it performs a ray tracing algorithm and the impulse response resolution was 1 ms.
analysis according to the image source method and As the late-part of the ray tracing algorithm can be
then, a diffuse reflection algorithm allows studying the considered as a Monte Carlo approximation, the
late decay of the sound (Dalenbäck 1996). The output of CATT-Acoustics may suffer from run-to-
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

software utilizes a randomized tail-corrected cone run fluctuations (Summers et al. 2005). This problem
tracing that combines features of both specular cone was reduced by using a large number of rays and a
tracing and standard ray tracing (Summers et al. 2005). truncation time longer than the reverberation time.
For diffuse-reflection order equal to one, a number Simulations were repeated in order to verify that
of elementary sources are created on each diffusing differences between simulation outputs were always
surface. Power is radiated from the surface according below 1% of their values (Vorländer 2008). This
to the Lambert’s cosine law and is proportional to demonstrated that both the truncation time and the
s(1 7 a ), where s is the scattering coefficient of the number of rays were correctly chosen.
surface and a is the absorption coefficient. Similarly, Figure 1 shows the main geometrical aspects of
the powers of first-order and second-order specular each simulated church. The location of the receivers
reflections are reduced by (1 7 s)(1 7 a). and the sources was parameterized as percentage of the
For the direct sound and first-order specular dimensions of the church. In this way, the internal
reflections, values at frequencies between the input organization of the churches, and the locations of
octave-band centre frequencies are found by interpola- receivers and sources were easily defined.
tion, using a cubic-spline method. For first-order An important aspect in the simulation was the
diffuse reflection, the software creates many secondary definition of the area occupied by pews. This area was
sources whose number is automatically decided pro- established considering the parameters that are taken
portionally to the scattering of the material. All the into account by the Vatican for new buildings (CEI
reflections both of the first-order and higher-order 2007). For example, according to this regulation, a
reflections are divided by octave-band filters before church with a volume of 2500 m3 should contain 3000
being summed together (Hodgson et al. 2008). worshipers. Consequently, the surface occupied by the
The room-prediction algorithm includes phase in pews was established.
order to predict impulse response. The reflection-path The simulations were performed with the source in
phase calculation method depends on the reflection the five different positions described in Section 2.1. The
order. For the direct sound and first-order specular organ was considered as an omnidirectional source
reflections, a minimum-phase response is considered. (Meyer 2003), with a white noise emission spectrum,
For first-order diffuse reflection and all second- and and equal sound pressure levels in all octave bands.
higher-order reflections, the phase is that of the linear The locations of the source were at a distance of 0.5 m
phase and finite impulse responses. Main advantages from near walls to have a better diffusion of the sound.
of CATT-Acoustics are that room surfaces can be The 10 positions of the receivers were:
modelled with frequency-dependent absorption and
scattering coefficients and that cone tracing is per- . one behind the altar at a height of 1.5 m, to
formed independently for each frequency band. More simulate the position of the celebrant;
details about the software are available in Hodgson . eight in area of the worshipers at a a height of
et al. (2008) and Dalenbäck (2008). 1.2 m, to simulate seated people;
CATT-Acoustics is similar to ODEON, the other . one in the back of the church at a height of
worldwide known software. Recent researches have 1.5 m, to simulate people standing in the back of
shown that the output of these programs is particularly the church.
similar (Bork 2005, Hodgson et al. 2008). In fact, they
use a mixed algorithm with the method of images and Table 2 records the absorption and scattering
tracing, and both take into account diffuse surface coefficients of the materials adopted in the simulations:
 8 U. Berardi

. the flooring was considered as a rigid surface surfaces combined with architectural complexity,
made of stones, with specular sound reflection mostly related to the presence of aisles, domes or
properties, and very low absorption; chapels, cause high scattering of sound. Finally, the
. the ceiling was considered lightly absorbent at values of the scattering coefficients also considered the
low frequency in order to simulate the typical effect of decorations (Berardi 2011).
effect of false ceilings; High scattering coefficients were assigned to the
. the lateral walls were considered lightly absor- upper and the lateral surfaces of pews. This approach
bent at high frequencies in order to take into was used in other simulations which have been
account the increasing absorption of plaster and calibrated with real measurements (Martellotta 2009).
gypsum with frequency; The scattering of the flooring material was considered
. the back wall was considered low absorbent with low, according to the supposed rigid surface. Conse-
a light increase of absorption at higher quently, scattering coefficients varying from 0.10 at
frequencies; 125 Hz to 0.20 at 4 kHz were assigned to the floor. The
. the area occupied by pews was simulated as linear increase of 0.02 includes the frequency depen-
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

highly absorbent which corresponded to the dence of scattering (Cox and D’Antonio 2004, Zeng
condition of full occupancy. The high absorption et al. 2006). Finally, the scattering coefficient of the
of clothes and the difficulty in modelling human ceiling and lateral walls was assumed to increase with
bodies made particularly difficult the simulation the frequency from 0.20 at 125 Hz up to 0.45 at 4 kHz.
of pews. Consequently, the area occupied by the This arrangement takes into account the effect of
people was simulated as a box, with a size equal decorations on these surfaces together with their
to that occupied by the pews (Martellotta 2009). simplicity in the simulated models (Martellotta 2009,
Moreover, recent studies showed that the sound Garrido et al. 2012).
absorption of the lateral and superior sides of
people is not the same (Beranek et al. 2001).
Different absorption values were hence used for 4. Simulation results
the lateral and upper surfaces of the pews. Simulations were performed for each of the 125
Finally, values of coefficient of absorption of configurations. A large amount of data was obtained
pews were taken by Martellotta and Cirillo as in each simulation many parameters were predicted.
(2009). Moreover, as any parameter was calculated in the
frequency bands from 63 to 4000 Hz in each of the 10
The scattering coefficient indicates the ratio between positions of the receiver, a large amount of data was
the part of the energy which is not reflected in a obtained.
specular way and the total reflected energy (Barron The simulation aimed to identify relationships
2003). Given the simple box shapes which were between the shape of the church and acoustic para-
simulated in this study, scattering values were lightly meters. Only the C80, JLF and Ts were considered. The
increased in the simulation to take into account the choice to avoid considering common parameters in
roughness of real surfaces. In particular, limited room acoustics, such as the reverberation time or the
dimensions of a surface as well as edge diffractions early decay time, was related to the fact that the use of
can cause diffractions (Cox et al. 2006, Zeng et al. the same materials in the simulated churches resulted
2006). Martellotta (2009) found that richly decorated with inconsistent differences of the reverberation time

Table 2. Material properties selected for the simulation.

Absorption coefficient Scattering coefficient****

125 Hz 250 Hz 500 Hz 1 kHz 2 kHz 4 kHz 125 Hz 250 Hz 500 Hz 1 kHz 2 kHz 4 kHz
Floor* 0.01 0.01 0.02 0.02 0.03 0.03 0.10 0.12 0.14 0.16 0.18 0.20
Roof** 0.20 0.15 0.10 0.08 0.07 0.07 0.20 0.25 0.30 0.35 0.40 0.45
Lateral walls* 0.06 0.06 0.06 0.08 0.09 0.10 0.20 0.25 0.30 0.35 0.40 0.45
Back wall* 0.03 0.05 0.06 0.08 0.09 0.10 0.20 0.25 0.30 0.35 0.40 0.45
Upper surface 0.50 0.60 0.70 0.80 0.80 0.80 0.50 0.55 0.60 0.65 0.70 0.75
of pews***
Lateral surface 0.05 0.10 0.27 0.27 0.14 0.17 0.50 0.55 0.60 0.65 0.70 0.75
of pews***

Note: Data sources for the absorption coefficient: *Vorländer, 2008; **Barron, 1993; ***Martellotta and Cirillo, 2009. Data source for scattering
coefficients: ****Martellotta, 2009.
 Journal of Building Performance Simulation 9

between churches with the same volume. Conse- were obtained for the reflections on lateral walls. In
quently, the investigation of the differences of the general, the longitudinal typology with the source on
reverberation time was of little importance. However, the main axis (source at positions A or E) created
the results of the reverberation time were used to acceptable acoustical conditions, except for the largest
compare simulation output with prediction according volume (50,000 m3). In this large volume, the values of
to Sabine’s formula. This comparison was an indirect JLF with source at A and E were 16.2% and 16.3%,
way to test the robustness of the simulations and the respectively.
agreement between CATT-Acoustics results and ex- Considering another typology, such as the
pectations with the theory of sound diffuse field. The stretched transversal typology, the highest results of
limit for the comparison was the just notable difference JLF were obtained with the source in the position C.
(JND) of the reverberation time, for which a value of However, this provoked an inhomogeneous distribu-
5% of the reverberation time is generally accepted. The tion of values (Figure 5). In this regard, a dissymme-
difference between simulated reverberation times and trical position of the source in a church with a
analytical calculated reverberation times was below 0.9 transversal plan, such as sources at B, C or D,
maximized the JLF in the smaller volumes (2500 m3),
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

JND for each of the 125 simulations. This result

matches the findings of other studies which confirmed when reflective surfaces surrounded the sound source.
the reliability of acoustic simulations in highly diffus- On the contrary, JLF in a stretched transversal plan
ing and geometrical simple spaces in which diffractions with a volume of 50,000 m3 rarely exceeded 15%.
and wave effects are negligible (Bork 2005, Hodgson Considering the result of JLF of stretched transversal
et al. 2008, Garrido et al. 2012). plan when the source was at C, it can be argued that
The figures below report the simulation results for high values of JLF occurred near the corners of the
the different parameters at 500 Hz. Middle frequency church, in positions which were not occupied by the
values seemed representative for the scope of this worshipers.
study, also considering that the 500 Hz values are Figure 6 shows the colour maps of JLF for different
recommended, in case of brevity, by the ISO 3382-1. typological plans with the sound source at the position
Figure 4 shows colour maps of JLF in longitudinal E. This figure allows us to compare the lateral energy
churches when the source is located in front of the altar arriving at the receivers in different plans when the
(at position A). All the colour maps presented in this source is behind the altar. This position is common in
paper were realized for a grid of 1 6 1 m. Considering both traditional and new buildings because it has the
the dependence of the values of JLF from the volume of advantage to allow an easy perception of the sound
the church, Figure 4 shows the simulation results for source, which is in line with the altar. The longitudinal
two volumes, 10,000 and 50,000 m3. As expected, the plan showed JLF values above 20% among the
values of JLF in the smaller church are higher than receivers with a general homogeneity among all the
those of the larger ones. Figure 4 shows that positions of the pews. Considering the square and
satisfactory values of JLF in the areas occupied by transversal plans, high differences among the receivers
the worshipers were obtained within a longitudinal were obtained. In fact, the colour maps show that the
shaped church with a volume of 10,000 m3. In fact, JLF pews which are on the sides of the altar had JLF values
often reached value above 20%. These high values below 20%. This problem was also more critical in the

Figure 4. Colour maps of JLF in % at 500 Hz for the volume of 10,000 m3 (left) and 50,000 m3 (right) in longitudinal churches
with the sound source at position A.
 10 U. Berardi
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

Figure 5. Colour maps of JLF in % at 500 Hz for the volume of 2500 m3 (left), 10,000 m3 (centre) and 50,000 m3 (right) in
churches of stretched transversal typology with source location at C.

Figure 6. Colour maps of JLF in % at 500 Hz for the volume of 10,000 m3 in stretch longitudinal, square, stretch transversal
and transversal churches with source location at E.

stretch transversal typology where the JLF values The Ts allows us to discuss the time the sound
resulted below 20% also in the positions in front to energy takes to reach the receiver. This parameter is
the altar. This shows that in case of transversal significantly affected by the location of measurement in
typologies, the source behind the altar has to be relation to the source, as it is evident in Figures 7 and
discouraged, or alternatively, systems to increase 8. Positions closer to the source generally show a lower
lateral reflections have to be designed. value of Ts. For example, results of the Ts in the
 Journal of Building Performance Simulation 11

stretched longitudinal plans, with the source at A or E, source at the position E. This figure, together with
showed a gradual growth of the value of Ts with the Figure 7, allows us to compare the centre time when
distance. the source is behind the altar. Values of Ts with a
The typology with the lowest Ts values was the longitudinal plan rarely exceed 250 ms, for churches
square. In this typology, the distance between source no longer than 40 m (below 20,000 m3). This is evident
and receiver was minimized on average. Consequently, in Figure 8, where the Ts assumed values from 140 to
the value of Ts was smaller than in other typologies, 180 ms according to the distance of pews from the
and it generally resulted as satisfactory. Figure 7 shows altar. A homogeneity of values of Ts also resulted for
the Ts maps in a square church of 10,000 m3 with the stretch longitudinal typology which showed very
sources at different location. The maps show that if the low values among the receivers. In particular, the area
sound source is near to the pews, the Ts assumes around the altar (before the pews) showed values
satisfactory value in all the plan. In particular, source below 140 ms, whereas the pews resulted in a strong
at B resulted in an absence of Ts values above 260 ms homogeneity of Ts with values of 210 ms.
in the positions of pews. Also positions in D or E can Colour maps of C80 have not been shown for
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

be considered particularly adapted for this plan. brevity; however, the reader should consider that Ts
Obviously, also for the squared shape, Ts increased colour maps allow to have an idea of the behaviour of
with the volume of the church, assuming higher values the clarity in each room. Figure 9 shows the average
in churches of larger volumes. By contrast, the values of C80, JLF and Ts for different volumes and
stretched longitudinal typology determined the max- typological plans with the sound source at the
imum distance between source and receiver. Values of positions A, D and E. The choice of these source
Ts higher than 300 ms were obtained in churches with positions was suggested by the interest for comparing
a volume of 10,000 m3 only, whereas Ts reached values values in positions on the symmetrical axis of the
as high as 660 ms with volumes of 50,000 m3. church, which corresponded to a typical historical
Figure 8 shows the colour maps of Ts for the configuration of the church, and in a position which
longitudinal and transversal plans with the sound better agrees to the Second Vatican Council (CEI

Figure 7. Colour maps of Ts in ms at 500 Hz for the volume of 10,000 m3 in square churches with source locations at B, C, D
and E.
 12 U. Berardi
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

Figure 8. Colour maps of Ts in ms at 500 Hz for the volume of 10,000 m3 in longitudinal and stretch transversal churches with
source location at E.

2007). The graphs in Figure 9 show that the acoustical revealed that the average Ts was 123 ms, as a
conditions within the churches were particularly consequence of the very short distance between the
influenced by the typology of the church whenever source and worshipers. Such a low distance provoked a
the volume was larger than 10,000 m3. In fact, smaller very good intelligibility of the source, but a poor
volumes often showed similar values among different quantity of sound energy from lateral directions.
typologies. Considering the churches with a volume of
The values of C80 in churches of large volume were 50,000 m3, the transversal plan was the shape which
particularly critical, being 7 5 dB on average (with guaranteed the lowest value of Ts, independently from
values of 7 10 dB at some receivers). The source at A the position of the source. This arrangement showed
position gave higher values of C80 than the source at the smallest values of Ts: 263 ms with source at A,
D. However, this was not the case for the stretched 275 ms at B, 260 ms at C, 267 ms at D and 256 ms at
transversal typology. The singular behaviour of the E. For all the other volumes and plans, Ts always
simulations for the stretched transversal typology can resulted below 260 ms. Finally, the analysis of Ts
be attributed to the late arrival (after 80 ms) of showed that only for the larger volumes, the source
reflections from lateral walls. should be positioned in a way to reduce the distance
Churches with smaller volumes (2500 and 5000 m3) from the worshipers.
showed good values of JLF for every typology. In A further investigation was performed to under-
particular, it emerged that the JLF was higher when the stand the relationship between parameters and dimen-
source was in position A than in D. This shows that a sional ratios of churches. In literature, some
traditional location of the organ (position A) resulted relationships exist between the width of a room and
in higher values of JLF than locations more common in its JLF (Klosak and Gade 2008). These relationships
new buildings, such as the position D. have been developed for concert halls, but they are not
The worst condition for the lateral effects of the really useful in churches (Carvalho et al. 1997, Cirillo
sound was given by the stretched transversal typology and Martellotta 2006, Berardi et al. 2009).
with the highest volume (50,000 m3), where average First of all, the results of the simulations were
values of JLF were low (7.6%, 10.5% and 8.6% with plotted against the width of the room, but no trends
the source in A, D and E, respectively). In these cases, could be recognized. As changes in height are known
the simulations showed that the source should be in to have some influence over JLF (Klosak and Gade
other locations to obtain sufficient results (JLF was 2008), other geometrical dimensions were hence
21.7% and 18.1% with the source at B and C, considered. In particular, the results of C80, JLF and
respectively). Ts were grouped according to the length to width ratio
Figure 9 also shows that the JLF for the transversal of the church. Figure 10 shows the results when the
plan with the source at D is particularly low for a source is in A (for every plan), D (for the stretched
volume of 20,000 m3. A further analysis of this case longitudinal, square and stretched transversal plans)
 Journal of Building Performance Simulation 13
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

Figure 10. Relation between room width and simulation

results of C80, JLF and Ts with the sound source in different
positions (in A for every plan, in D for the stretched
longitudinal, square and stretched transversal plans and in E
for the stretched transversal plan. Data are shown together
with trends for source at A according to different the length-
to-width ratios of the church.

Figure 9. Average values of C80, JLF and Ts in churches of analysis, R2 was calculated by the square of the sample
different shapes and volumes with the sound source in
position A (solid lines), D (dashed lines) and E (dot-dashed
correlation coefficient between the simulation results
lines). and their predicted values according to the fitting line.
The results of R2 were high for every regression and
showed that each parameter was well related to both
and E (for the stretched transversal plan). The figure the width and length of the church. The formulae
only reports the average values among the receivers. recorded in Figure 10 help to estimate each parameter
Moreover, three regression lines for each parameter (C80, JLF, Ts) in box-shaped churches according to
are given for different typologies of churches: stretched their width and length.
longitudinal, square and stretched transversal (with Figure 10 shows that an increase of width reduces
ratios of 2:1, 1:1, 1:2, respectively). the C80 in churches with the same length to width ratio.
The coefficient of determination R2 was considered The regression lines show that C80 is strongly related to
to assess the accuracy of the correlation. In the present the ratio between the floor dimensions of the church.
 14 U. Berardi

Ts (ms)
Table 3. Measured and predicted values of C80, JLF, and Ts in five existing churches with a rectangular shape and different length-to-width ratios (measured data in Cirillo

277
369
205
328
472
Figure 10 also shows that the reduction of JLF with the
increase of the width is smaller than the corresponding

Predicted values
reduction of C80 in churches with the same length to

Ts (ms) C80 (dB) JLF (%)

width ratio.

21.4
19.6
22.4
20.4
9.2
A similar behaviour concerns Ts, which seems to be
significantly influenced by the length to width ratio.
The results in Figure 10 show that Ts significantly

72.4
74.2
0.6
73.4
73.6
increases with the room width. Moreover, the width of
the room has a larger influence over the average Ts in
the stretched longitudinal typology than in the
stretched transversal plan. This behaviour can be

*
*
*
*
*
related to the greater increase of the distance between

Measured values
the source and receivers for the stretched longitudinal

C80 (dB) JLF (%)

plan than in the transversal plan. The regression lines

23.8
20.0
24.1
19.9
10.6
of Ts had higher R2 coefficient than for JLF. Moreover,
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

the distance between the regression lines of Ts suggests

that it is particularly important to know the plan of the

7 5.3
7 8.3
7 2.3
7 6.8
7 5.1
church in order to understand which values the Ts
could assume. The proposed formulae allow us to
make a preliminary estimation of Ts given the width of

Width (m) Ratio

L/W

2:1
2:1
1:1
2:1
1:2
the room and its length.
Lastly, the relationships in Figure 10 were applied
to some real buildings in which acoustical parameters
had been measured previously. Historical churches

29
38
22
34
75
often differed from simple box-shaped. In fact, the
architectural and theological reasons behind the shape
of a church often discouraged rectangular shapes over
Volume Length
(m)

55
67
22
58
56
the centuries. However, five churches were selected, in
a way to compare different periods (from Early-
Christian churches to modern churches), volumes
22,500
23,000
19,000
9,000
50,000
(m3)

and length-to-width ratios. Table 3 shows the mea-

sured values for C80 and JLF in these churches (Cirillo
and Martellotta 2006). Details of the measurement
T (s)

3.9
6.0
8.9
4.4
techniques can be found in previous publications and 6.1
are not reported here in the interest of brevity.
2004/Contemporary

Measurements of Ts were not available. Moreover, it

1099/Romanesque
1508/Renaissance

should be noted that the measurements were taken

539/Byzantine

1970/Modern

considering a source position different from that used

Period/Style

for the simulation study and hence, higher errors in the

estimation could be expected.
and Martellotta, 2006 and Berardi et al., 2009)

The comparison between the values obtained with

the formulae in Figure 10 andthe measured values
Santa Maria della Consolazione (Todi)

shows good agreement, especially for the JLF. The

Sant’Apollinare in Classe (Ravenna)

average rms errors between estimation and measure-

Basilica of Sant Ambrogio (Milan)

ments were 0.85 dB for C80 and 2.18% for JLF. These
The Concattedrale of Taranto

results are remarkable considering the different materi-

Padre Pio Pilgrimage Church
(San Giovanni Rotondo)

als, shapes and decorations of the real churches, and the

difference of these churches from the simple box-shaped
churches which were used to obtain the formulae.
In particular, the church with the largest errors in
the predictions was Basilica of Sant’Ambrogio (Mi-
lan). This church is an asymmetrical room with high
cross vaults finished in smooth plaster. The high
Church

reverberation time of this church, that is 6.0 s, is

among the reasons of the large errors in the
 Journal of Building Performance Simulation 15

predictions. In fact, late and strong reflections from the important buildings for worship and teaching pur-
high vaults provoked particularly low values of C80. poses, but little consideration is taken regarding their
However, the early lateral energy is well predicted with acoustic design. This paper has presented and com-
the formulae proposed in this paper. pared simulation results of churches of different
Prediction values for the Padre Pio Pilgrimage typologies, volume and internal organization.
Church strongly agree with measured values. This Some of the most diffused typologies of churches
church has an organizational plan that it is very similar have been defined, and different internal floor plans
to that of a stretched transversal typology, although it have been considered. The study has focused on three
has a semicircle shape. This shape differs from the parameters which represent different spatial qualities.
objective of this study which focused on rectangular The results have been analyzed averaging the values
shapes. However, the shorter distances of the back assumed by these parameters in 10 positions in each
walls in the Padre Pio Pilgrimage Church than in a church.
church with a rectangular shape of the same length and The square typology was the typology which
width did not prevent to estimate the acoustical recorded the lowest Ts. A square church has showed
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

parameters, both the ratio between early and late high values of JLF. Differently, the stretched long-
energy and the amount of lateral reflections. The good itudinal typology showed high values of JLF, but also
agreement suggests that the formulae in Figure 10 may high values of Ts. This may affect the intelligibility in
be interesting in churches with a shape different from a the church.
rectangular plan. The results showed that acoustic conditions are
In general, the comparison between measured and particularly influenced by the typology of the church
predicted values in Table 3 suggests that the formulae only when the volume is greater than 10,000 m3. In
proposed in this paper are well suited for the estimation churches of small volumes, the parameters showed
of JLF, which shows to be particularly dependent on the similar results for different typologies and internal
width of the church. However, the measured values of organizations. Obviously, this could result by the
C80 fall down the prediction of the formulae. This selection of the same materials among simulations. In
confirms a result already known in the studies about the fact, the use of the same absorption and scattering
energy decay in churches which have been reviewed in coefficients for different churches influenced the
Section 1. In fact, the ‘disproportionate’ nature of real similarities of results in smaller churches.
churches and architectural elements such as side aisles, The simulations have shown a lower decrease of
chapels, vaults and domes which characterized most of JLF and a significant increase of C80 and Ts with the
the real buildings provoke the scattering of the sound, room width. Moreover, the length-to-width ratio (and
and a late arrival of the first reflections. This justifica- hence the plan of the church) has shown a significant
tion agrees with the better prediction of JLF respect to influence over the values of both C80 and Ts.
C80. Moreover, this justifies the better prediction in The paper has shown preliminary results of a
modern churches, where architectural decorative ele- typological study and has established simple relation-
ments are less frequent. ships to quickly evaluate some acoustic parameters of a
Finally, it has to be stated that the validity of the church. The potentialities of these relationships should
formulae is limited to the assumptions done for the be considered useful to make preliminary estimations of
simulations. In particular, the material properties parameters other than the reverberation time in
influence the acoustics of any room, and hence the churches.
formulae may not be generalized to box-shaped Future studies are needed to investigate further the
churches with materials and geometrical characteristics effect of different room shapes and materials. Finally,
different from the input data in the simulations. This the relationships proposed in this study have to be
means that the formulae would give considerable errors validated in a larger sampling of churches.
if the rectangular room is not a church or its acoustical
properties significantly differ from the values in Table 2.
In particular, the pews, which represent a typical References
absorbent surface of churches, are a distinctive char- Barron, M., 1993. Auditorium acoustics and architectural
acteristic of rooms where the formulae are valid. design. London: E & FN Spon.
Barron, M. and Lee, L., 1998. Energy relations in concert
auditoriums. Journal of the Acoustical Society of Amer-
ica, 84 (2), 619–629.
5. Conclusions Beranek, L., Nishihara, N., and Hidaka, T., 2001. Mechan-
ism of sound absorption by seated audience in halls.
This research has studied the acoustics of box-shaped Journal of the Acoustical Society of America, 110 (5),
churches through software simulations. Churches are 2400–2411.
 16 U. Berardi

Berardi, U., 2011. Predicción de las variaciones espaciales en Hodgson, M., et al., 2008. Comparison of predicted,
la acústica de diferentes tipologı́as de Iglesias. Revista de measured and auralized sound fields with respect to
Acústica, 42 (3–4), 8–16. speech intelligibility in classrooms using CATT-acoustic
Berardi, U., 2012. A double synthetic index to evaluate the and ODEON. Acta Acustica United With Acustica, 94
acoustics of churches. Archives of Acoustics, 37 (4), 521– (6), 883–890.
528. ISO 3382-1, 2009. Acoustics – measurement of room acoustic
Berardi, U., Cirillo, E., and Martellotta, F., 2009. A parameters - part 1: performance spaces. Geneva: Inter-
comparative analysis of energy models in churches. national Standard Organisation.
Journal of the Acoustical Society of America, 126 (4), Klosak, A.K. and Gade, A.C., 2008. Relationship between
1838–1849. room shape and acoustics of rectangular concert halls.
Bork, I., 2005. Report on the 3rd round robin on room Journal of the Acoustical Society of America, 123 (5),
acoustical computer simulation-part II: calculations. 3199.
Acta Acustica United With Acustica, 91 (4), 753–763. Magrini, A. and Ricciardi, P., 2003. Churches as auditoria:
Carvalho, A.P.O., 1994. Influence of architectural features analysis of acoustical parameters for a better under-
and styles on various acoustical measures in churches. standing of sound quality. Journal of Building Acoustics,
Dissertation (PhD). University of Florida. 10 (2), 135–157.
Carvalho, A.P.O., Morgado, A.E.J., and Luis Henrique L., Marshall, L.G., 1994. An acoustics measurement program
Downloaded by [Memorial University of Newfoundland] at 01:55 18 July 2014

1997. Relationships between subjective and objective for evaluating auditoriums based on the early/late sound
acoustical measures in churches. Journal of Building energy ratio. Journal of the Acoustical Society of America,
Acoustics, 4 (1), 1–20. 96 (4), 2251–2261.
CEI (Commissione Episcopale Italiana), 2007. Parametri Martellotta, F., 2008. Subjective study of preferred listening
indicative per l’edilizia di culto (in Italian). Rome: Servizio conditions in Italian Catholic churches. Journal of Sound
nazionale per l’edilizia di culto. and Vibration, 317 (1–2), 378–399.
Cirillo, E. and Martellotta, F., 2005. Sound propagation and Martellotta, F., 2009. Identifying acoustical coupling by
energy relations in Churches. Journal of the Acoustical measurements and prediction-models for St. Peter’s
Society of America, 118 (1), 232–248. Basilica in Rome. Journal of the Acoustical Society of
Cirillo, E. and Martellotta, F., 2006. Worship, acoustics and America, 126 (3), 1175–1185.
architecture. Essex: Multi-science Publishing. Martellotta, F. and Cirillo E., 2009. Experimental studies of
Cox, T.J. and D’Antonio, P., 2004. Acoustic absorbers and sound absorption by church pews. Applied Acoustics, 70
diffusers. New York: Spon Press. (3), 441–449.
Cox, T.J., et al., 2006. A tutorial on scattering and diffusion Martellotta, F., et al., 2009. Guidelines for acoustical
coefficients for room acoustic surfaces. Acta Acustica measurements in churches. Applied Acoustics, 70 (2),
United With Acustica, 92 (1), 1–15. 378–388.
Dalenbäck, B.I., 1996. Room acoustic prediction based on a Meyer, J. 2003. Kirchenakustik. Frankfurt am Main: Verlag
unified treatment of diffuse and specular reflection. Erwin Bochinsky.
Journal of the Acoustical Society of America. 100 (2), Segura, J., et al., 2011. A comparison of different techniques
899–909. for simulating and measuring acoustic parameters in a
Dalenbäck, B.I., 2008. CATT-Acoustic v8 user’s manual. place of worship: Sant Jaume Bası́lica in Valencia, Spain.
Gothenburg, Sweden: CATT. Acta Acustica United With Acustica, 97 (1), 155–170.
de Sant’Ana, D.Q. and Zannin, T.P.H., 2011. Acoustic Sendra, J., Zamarrenno, T., and Navarro, A., 1999. Acoustics
evaluation of a contemporary church based on in situ in churches. Computational Acoustics in architecture.
measurements of reverberation time definition and Southampton: Computational Mechanics Publications.
computer-predicted speech transmission index. Building Soeta, Y., et al., 2012. Effects of sound source location and
and Environment, 46 (2), 511–517. direction on acoustic parameters in Japanese churches.
Desarnaulds, V., 2002. De l’acoustique des e´glises en Suisse – Journal of the Acoustical Society of America, 131 (2),
Une approche pluridisciplinaire. Thesis (PhD). Ecole 1206–1220.
Politechnique Federale de Lausanne. Strappa, G., 2005. Edilizia per il culto. Turin: Utet-Scienze
Engel, Z. and Kosala, K., 2005. Acoustic properties of Tecniche.
the selected churches in Poland. Mechanics, 24 (3), 173– Summers, J.E., et al., 2005. Adapting a randomized beam-
181. axis-tracing algorithm to modeling of coupled rooms via
Fasold, W. and Veres, E., 2003. Schallschutz und raumakustik late-part ray tracing. Journal of the Acoustical Society of
in der praxis. Planungsbeispiele und konstruktive lösungen. America, 118 (3), 1491–1502.
Auflage: Beuth. Vorländer, M., 2008. Auralization. Berlin: Springer-Verlag.
Galindo, M., Zamarreno, T., and Girón, S., 2005. Acoustic Weitze, C., et al., 2002. The acoustical history of Hagia
analysis in Mudejar-Gothic churches: experimental Sophia revived through computer simulation. In: Pro-
results. Journal of the Acoustical Society of America, ceedings of the forum acousticum, 16-20 September,
117 (5), 2873–2888. Seville, Spain.
Galindo, M., Zamarreno, T., and Girón, S., 2009. Acoustic Zeng, X., Christensen, C.L., and Rindel, J.H., 2006. Practical
simulations of Mudejar-Gothic churches. Journal of the methods to define scattering coefficients in a room
Acoustical Society of America, 126 (3), 1207–1218. acoustics computer model. Applied Acoustics, 67 (8),
Garrido, J.A., Zamarreño, T., and Girón, S., 2012, Virtual 771–786.
models for the prediction of acoustic fields of Manuel de
Falla Auditorium in Granada, Spain. Applied Acoustics,
73 (9), 921–935.