| Abstrakt: |
Traditional Sabine equations still are used for factories or offices where diffuse sound fields rarely occur and prediction can be inaccurate. More recently, methods based on geometric acoustics have been developed which require large computing time and which demand better defined input data. A problem, often encountered, is how to include input data which is appropriate, accurate and relatively easy to obtain. Three acoustic models of a furnished room were created: a diffuse field, an image source and a ray tracing model. The initial values of absorption coefficient and sound power level were obtained by standard measurements and the sound propagation SP was predicted and compared with measurement for each model. Then, the models were calibrated by altering the input parameters in order to minimise the difference between predicted and measured values. Sound pressure level due to two sources was also predicted and compared with measurement.For the room studied, the precision of the predictions, after calibration, is similar for the three models considered, with an average difference between simulated and measured values of less than 2 dB. Without the calibration procedure, the ray-tracing model gave the most precise first estimate. The diffuse and image source models needed significant modification of the input data to obtain a similar precision. The sound field in the room chosen for this study was nearly diffuse and simulation, based on geometric acoustics, did not offer clear advantages. However, this will not be the case for rooms with more complicated geometrical and acoustic characteristics such as in factories and offices. In addition, the image source model will not be appropriate for internal fittings which are much more complex than in the present study and an appropriate estimate of the scattering cross-section is problematical. In the ray tracing model, this problem is circumvented by incorporating the fittings as part of the geometry of the room. |
