CIPA 2003 XIX th International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
232 
2. SIMULATIONS 
Before the room acoustic computer models were developed the 
best way to make a prediction for room acoustics was through 
measurements in a scale model. Nowadays the room acoustic 
computer model has reached a good level of reliability (Rindel, 
2000; Lynge Christensen, 2001). Modifications can be made in 
an easy way and in a short time and the results can be evaluated 
by listening too. Different softwares are available for room 
acoustic simulations. In this study the ODEON (Odeon 
websites) software package is used. 
2.1 The simulation principles 
For the acoustic simulation, a three-dimensional model has to 
be created (Fig.2). Materials with their absorptions coefficient 
and scattering coefficient have to be assigned to each surface in 
the model in order to obtain the desired acoustical behaviour. 
When the simulation is based on an existing room, the materials 
have to be assigned so that the model behaves like the real 
measured room. This procedure is called calibration. Then 
source and receiver positions have to be fixed. For the 
simulation algorithm two classical geometrical methods, Ray 
Tracing and Image Source Method, are combined for early 
reflections. For the late reflections a hybrid between Ray 
Tracing and Radiosity is used: a large number of rays are 
emitted in all the directions from a source point. The rays are 
traced around the room, loosing energy at each reflection 
depending on the surface’s characteristics. At the same time a 
secondary source is generated every time a ray hits a surface. Its 
energy is a small portion of the primary source. The results at 
the receiver positions are obtained by the reflections collected at 
the receiver position. Through the impulse response, calculated 
in each receiver position, the acoustics parameters are obtained 
and shown in octave bands. 
2.2 The simulation procedure 
The calibrated room acoustic model of the Royal Theatre was 
used for all the simulations. Because of its symmetry just one 
side of the theatre was tested. The directional source was placed 
on the stage in three different positions: the first position at 1 m 
to the symmetry axis and 1 m to the fire curtain, the second 
position at 1 m to the symmetric axis and 5 m to the fire curtain 
and the third position at 4 m to the symmetric axis and 1 m to 
the fire curtain, all of them at 1.5 m above the stage floor. These 
positions were chosen as a singer’s significant positions. The 
omnidirectional source was placed in the pit in three different 
positions as well: the first violin position, the oboe position and 
the brass position, all of them at 1.2 m above the pit floor. The 
directional and the omnidirectional sources were simulated with 
the same power level in each octave band in order to be 
comparable. The sound pressure level simulated by the software 
at each receiver position was the acoustic parameter used in this 
study. 
The energetic average of the sound pressure level obtained at 
each receiver position (8 in the stalls and 13 in all the balconies) 
for each single source, playing one at a time in the three 
different positions on the stage and in the pit was calculated. 
The balance was calculated as the difference between the 
energetic average of the sources on the stage and those in the 
pit. The results are showed as an average in two frequency 
band, each covering two octaves: 500 - 1000 Hz and 2000 - 
4000 Hz. This process was repeated for each simulation. 
2.3 Acoustical design elements 
Different changes were carried out based on geometries and 
materials. The first attempt was to investigate the pit. This 
element is historically important, but can be modified without 
changing the identity of the theatre. 
The pit depth was changed in order to see its influence on the 
balance. Firstly a simulation was performed with a pit floor 
depth level of 1.5 m from the edge of the barrier which 
separates the pit and the stalls. In the results this change is 
referred to as “1.5 m”. The second attempt was simulated with a 
pit floor depth level of 2 m from the edge of the barrier. In the 
results this change is referred to as “2 m”. The pit floor was 
fixed at 2 m from the edge of the barrier in all the following 
changes. 
Subsequently the influence of changing the materials of the pit 
walls was evaluated. Changes of the absorption coefficient of 
the wall in the back of the pit, were tested. The new material 
chosen was more absorbent than the previous one. The purpose 
was not to look for a particular kind of material, but to search 
for elements which could influence the balance. In the results 
this change is referred to as “Pit back wall”. This material was 
fixed for following simulations. 
The covering of the pit fence facing the musicians was changed 
too. A more absorbent material was chosen. In the results this 
change is referred to as “Barrier”. 
Other elements were tested. The barrier’s height and the stage’s 
slope at 5%, 8% and 10%, but no significant aspect on the 
balance were detected. Because of this these results are not 
shown. 
2.4 Results 
The frequencies of main interest in this study are the octave 
bands 2000 and 4000 Hz, where the formant of the singer’s 
voice is located and where the emission spectrum of the voice 
can be compared with that of the orchestra (Sundberg, 1977; 
Meyer, 1986). In order to have a more complete overview, 500 
and 1000 Hz have been considered too. 
It seems that two different balances exist in an opera house: one 
for the stalls and another for the balconies. That’s the reason 
why the results have been divided by stall and balconies: two 
completely different behaviours are observed. The results are 
shown based on the distance between source and receiver 
positions. 
Figure 3 shows the calculated balance, obtained from the 
simulations. Changing the pit level from 1.5 m to 2 m the 
influence is noted in the stalls but nothing changes in the 
balconies, for both the average frequencies considered. Keeping