On the accuracy of models for predicting sound propagation in fitted rooms

The objective of this article is to make a contribution to the evaluation of the accuracy and applicability of models for predicting the sound propagation in fitted rooms such as factories, classrooms, and offices. The models studied are 1:50 scale models; the method-of-image models of Jovicic, Lindqvist, Hodgson, Kurze, and of Lemire and Nicolas; the empirical formula of Friberg; and Ondet and Barbry's ray- tracing model. Sound propagation predictions by the analytic models are compared with the results of sound propagation measurements in a 1:50 scale model and in a warehouse, both containing various densities of approximately isotropically distributed, rectangular- parallelepipedic fittings. The results indicate that the models of Friberg and of Lemire and Nicolas are fundamentally incorrect. While more generally applicable versions exist, the versions of the models of Jovicic and Kurze studied here are found to be of limited applicability since they ignore vertical-wall reflections. The Hodgson and Lindqvist models appear to be accurate in certain limited cases. This preliminary study found the ray-tracing model of Ondet and Barbry to be the most accurate of all the cases studied. Furthermore, it has the necessary flexibility with respect to room geometry, surface- absorption distribution, and fitting distribution. It appears to be the model with the greatest applicability to fitted-room sound propagation predictionCet article vise \ue0 aider \ue0 \ue9valuer la pr\ue9cision et l'applicabilit\ue9 des mod\ue8les de pr\ue9vision de la propagation du son dans les locaux am\ue9nag\ue9s comme les usines, les salles de classe et les bureaux. Les mod\ue8les \ue9tudi\ue9s sont des mod\ue8les au 1/50, les mod\ue8les de Jovicic, Lindqvist, Hodgson, Kurze, et de Lemire et Nicolas, bas\ue9s sur la m\ue9thode des images, la formule empirique de Friberg, et le mod\ue8le de tracage de rayons d'Ondet et Barbry. Les pr\ue9visions des mod\ue8les analytiques de propagation du son sont compar\ue9es aux r\ue9sultats des mesures de propagation du son dans un mod\ue8le au 1/50 et dans un entrep\uf4t contenant chacun des parall\ue9l\ue9pip\ue8 des rectangles de densit\ue9 variable ayant une r\ue9partition \ue0 peu pr\ue9s isotrope. Les r\ue9sultats indiquent que les mod\ue9les de Friberg et de Lemire et Nicolas sont fondamentalement incorrects. M\ueame s'il existe des versions plus universelles, les versions des mod\ue9les de Jovicic et Kurze, tels qu'\ue9tudi\ue9 es ici, s'av\ue9rent d'une applicabilit\ue9 limit\ue9e parce qu'elles ne tiennent pas compte des r\ue9flexions sur les parois verticales. Les mod\ue9les de Hodgson et Lindqvist semblent pr\ue9cis dans certains cas limit\ue9s. Cette \ue9tude pr\ue9liminaire a r\ue9v\ue9l\ue9 que le mod\ue9le de tracage de rayons d'Ondet et Barbry est le plus pr\ue9cis de tous ceux qui ont \ue9t\ue9 examin\ue9s. En outre, il est assez souple aux plans de la g\ue9om\ue9trie du local, de la r\ue9partition des surfaces absorbantes et de celle du mobilier. Il semble \ueatre le mod\ue9le qui se pr\ueate le mieux \ue0 la pr\ue9vision de la propagation du son dans les locaux am\ue9nag\ue9s.Peer reviewed: YesNRC publication: Ye