DEVELOPMENT OF A RAY TRACING COMPUTER MODEL FOR THE PREDICTION OF THE SOUND FIELD IN LONG ENCLOSURES

A ray tracing computer model, NORMAL, has been developed for predicting the sound field in any shaped enclosed space without curved surfaces, including long enclosures of rectangular cross-section. The model was developed specifically as a basis for the modelling of speech intelligibility in underground stations. The model predicts sound propagation throughout a space, and the impulse response at defined receiver points from which various acoustic parameters for the assessment of speech intelligibility can be calculated. This paper describes the way in which NORMAL models the space, receivers and sources, and the mathematics involved in tracing a ray and calculating the energy contributed to a receiver. The model has been tested in two hypothetical spaces to show that it is capable of predicting the characteristics of the sound fields that would be expected both in a diffuse space, and in a long, non-diffuse, enclosure. The reverberation time tail compensation, used to compensate for the loss of later reflections in the predicted energy decay curve, is described. The effects of including this compensation method on predicted reverberation time in diffuse and non-diffuse spaces are discussed, together with the feasibility of applying this correction method to long enclosures.