SIMULATION OF CERTAIN ACOUSTIC PROPERTIES OF THE "KNEŽEV DVOR" IN DUBROVNIK Bojan Ivančević 1, Marjan Sikora 2, Kristian Jambrošić 1 1 FER, Unska 3, Zagreb, Croatia, bojan.ivancevic@fer.hr, kristian.jambrosic@fer.hr 2 Enter, Split, Croatia, marjan.sikora@enter-st.hr Abstract "Knežev dvor", one of the most famous old buildings in Dubrovnik, is often used for concert purposes. Its atrium has specific acoustic characteristics, which raises the need to investigate and simulate them. The most interesting feature of the atrium is a partly lack of the ceiling while the materials of which it is built are highly reflective. Accordingly, the hearing impression is very special because, despite of the space liveness, there are no strong late reflections. It is almost impossible to make desired acoustic measurements during a concert. Thus, new software for 3D simulation of sound pressure distribution in enclosed spaces was developed. 1. INTRODUCTION The well-known "Knežev dvor" ("Lord yard" in english) in Dubrovnik is often used for music events. The musicians are placed in the atrium and the public in the surrounding lofts and staircases. It is an acoustically very interesting space because of its specific architecture design and often a great difference in the number and position of musicians and public. As there is no roof (except the porch above one part of the atrium), drastic change in the disposition of the musicians are possible, even during the concert. This is especially true in the case of rain when the musicians "hide" beneath the portico. The atrium is usually full of people during the concert while it is mostly empty during the rehearsing time. These facts indicate that special attention should be paid to the acoustic characteristics, thus raising the need to make examination of these characteristics. Fig. 1. Plan-view of "Knežev dvor". 2. THE ATRIUM OF "KNEŽEV DVOR" The atrium is surrounded by stonewalls giving the space a good isolation of surrounding noise. As it is placed inside the old town core, there is also no traffic noise. Plans of the atrium are shown if fig. 1 and 2. The whole yard is built of stone. Materials of which the atrium is built are solely stone and hard lime plaster. The floor is made of glazed stone blocks, as same as the staircases, balustrades etc. Vertical walls are mostly smoothly plastered; some parts are even of stone. Fig. 2. North side-view of the atrium. The ceiling is the sky, mostly unclouded, which drastically influences the overall impression, including the acoustical as well...
The complex, layered structure of the building and numerous vignettes, mostly made in stone or plaster, will considerably contribute to the diffuseness of sound. The vignettes include columns of different dimensions, column capitols, stuccos, etc. During the concert, simple wooden seats are positioned in the atrium and on the galleries. There are usually less seats then visitors so they often stand on stairs. The music events can be divided into soloist (piano up to a quartet) and orchestra (up to symphonic orchestra). In the first case the music source is placed beside the main staircase (piano always), sometimes beneath the base of the fountain (guitar, guitar trio), and sometimes beneath the angel statue (at the entrance of the "Knežev dvor"). In the second case, the orchestra is placed on the floor of the atrium, as well as the public. A large change in the sound distribution happens in the case of rain when the musicians (especially the orchestra) hides beneath the overarched part of the atrium, and some instruments change position being placed closer to the reflective walls. 3. REVERBERATION TIME SIMULATION Simulation of the reverberation time has been made for an empty, half-filled and with public filled atrium. There are 290 sedentary places in the atrium, 30 more on the gallery. Usually 200 visitors more are standing, but when the concerts are very attractive, even more then 600 visitors can stand (up to 900!). It was supposed that a half-filled atrium equals 300 visitors, and a filled atrium about 500 visitors. For simulation purposes only, three materials (with corresponding coefficients of absorption α for different frequencies) are taken into consideration (table 1). This was possible due to a relatively small error when avoiding other materials and because of the decrease of simulation time. Table 1 Absorption coefficients for the materials used in the simulation. Frequency Stone, plaster Public Ceiling opening 125 0,02 0,54 1 250 0,02 0,66 1 500 0,03 0,78 1 1 k 0,04 0,85 1 2 k 0,04 0,83 1 4 k 0,04 0,75 1 The absorption coefficients for the public relate to the coefficients after Eyring, accordingly given for m 2 filled with people sitting. The results of the simulation are given in table 2. and fig. 3. Table 2 Simulation results for the reverberation time of an empty, half-filled and filled atrium in seconds. RT RT (half Frequency RT (filled) (empty) filled) 125 5,78 3,04 2,28 250 5,78 2,73 1,99 500 5,78 2,48 1,76 1000 4,74 2,16 1,55 2000 3,52 1,89 1,42 4000 2,62 1,66 1,32 RT (s) RT (empty) RT (half full) RT (full) 8 6 4 2 0 125 250 500 1000 2000 4000 Frequency (Hz) Fig. 3. Reverberation time simulation. 4. DISTRIBUTION OF SOUND PRESSURE SIMULATION The simulation of sound pressure distribution in the atrium of the Knežev dvor was done using the HEAD 3D software [1-2]. The simulation is based on the method of virtual sources considering the direct sound as well as the reflections. It was done using 3D Studio MAX programme because it permits in the same time the calculation and the visualization. The model was developed according to plans in the 1:25 and 1:50 measure. The 3D model is done in a simplified version consisting of 325 triangles, fig. 4. The model includes only the basement of the atrium because of its dominant role in sound pressure distribution, fig. 5. Upper parts of the atrium were neglected because of their minor influence in the sound distribution and because they would slow done the simulation considerably. The simulation parameters were as follow: - absorption in air: 0.02 db/m - reflection coefficient for stone: 0.9989 - speed of sound: 343 m/s The simulation considered the direct sound and the first reflections. The resolution was 100x100 points while the surface of simulation was at 1 m height.
The sound source was omnidirectional at 1 m height. The referent sound pressure level was set 0 db at 1 m distance from the source, thus all other values are calculated in relation to the referent one. The simulation was done for the sound source placed in two characteristic places: in front of the staircase to the main gallery (fig. 5-7), and beneath the main porch (fig. 8-10). Fig. 5. Simulation for 250 Hz. Fig. 4. The model scheme in 3D Studio MAX environment. Fig. 6. Simulation for 1000 Hz. Fig. 5. Rendered 3D model of the atrium, view from above. 4.1. Sound source in front of the staircase Fig. 5 shows the results of the simulation for 250 Hz. The highest sound pressure level is marked white, while lower values have a greyer colour. The highest SPL was 9.98 db, and the lowest SPL was -48.49 db. The highest SPL is around the sound source where the sound is amplified by the reflections from the ground and the walls (the reflected sound is in phase with the direct one). In other parts of the atrium the SPL rises and lowers according to the interaction of the direct and reflected sound. The least SPL values are in the corridor of the atrium entrance. Fig. 7. Simulation for 4000 Hz.
Fig. 6 and 7 represent the SPL distribution for 1 khz and 4 khz respectively. Similar results are obtained as for fig. 5. The differences occur because of the different wavelength thus making a different interferential pattern. 4.2. Sound source beneath the porch Fig. 8-10 show the results of the simulation for 250, 1000 and 4000 Hz respectively. The results are something different as for the first case because of the strong reflections from the porch, thus making a more various interferential pattern. Fig. 10. Simulation for 4000 Hz. 5. CONCLUSION Fig. 8. Simulation for 250 Hz. The results of the reverberation time simulation differ somewhat from the expected values. Namely, the subjective impression of the reverberation time is smaller then the one obtained by the simulation. The reason is probably a huge difussioness of sound in the atrium that is not taken into account in the simulation. Furthermore, because of the partly open ceiling there is a lack of a part of ceiling reflections, which gives an impression of a smaller reverberation time. The simulation of the sound pressure distribution was made by an approximation of the space shape without taking into account the first floor and the ceiling because of the simulation speed requirements. A relatively rough, but typical distribution is obtained. In the next simulation step, more attention will be paid to finer details, so the precision of the simulation will rise. ACKNOWLEDGEMENTS We want to thank the Archive of the Ministry of culture in Dubrovnik, the Dubrovnik Festival, and the Institute of history of art in Zagreb helping us to collect the needed plans and other important data about "Knežev dvor". 6. REFERENCES Fig. 9. Simulation for 1000 Hz. 1. Marjan Sikora, Hrvoje Domitrović "CODA COmputerizeD Auralization", Proceedings of 100th AES Convention, Preprint 4230, Copenhagen, 1996. 2. Marjan Sikora, "Simulation of 3D ultrasound field distribution inside a scull", M.Sc. Thesis, Faculty
of EE and Computing, University of Zagreb, 2000.