To suggest a prediction method for sound transmission loss of multi-layered infinite panels since usual concepts of random or field incidence in the sound transmission theory for these panels often end up with significant discrepancies from the experimental results, it is pursued what directional distribution of incident energy is appropriate in predicting the sound transmission loss of multi-layered panels that correlates well with the actual measured data in the typical reverberation chambers. In other words, the focus of this study is finding a weighting function to represent the directional distribution of incident energy on the specimen in an actual test environment. For this purpose, numerical simulations by using a ray tracing technique and the experiment for the directional energy density in terms of sound intensity are performed. By comparing experimental results with simulated ones, the directional distribution of incident energy on the specimen can be approximately expressed by the Gaussian distribution function with respect to the angle of incidence. This concept of the Gaussian directional function is applied to the prediction of the acoustic performance for a variety of layered panel configurations as well as single panels. The compared results between the measurement and the prediction show good agreements that validate the use of this directional energy concept.
For analytical methods, this paper deals with the analytical models of sound transmission loss for the fire-resistive panel which is composed of mineral wools with high stiffness. From the comparison of the measurements with the predictions by using both the poro-elastic and the sandwich models, it is concluded that for double panels having the core with stiffness comparable to air stiffness, the accuracy of the predictions mainly depends on how to evaluate the mass-spring-mass resonance. As a result, the core stiffness is a key factor for selecting an analytical model. For the very stiff cores, however, both models result in good agreements with the measurement.
