Predicting noise in industrial workrooms using empirical models

Abstract

Sound fields in complex industrial workrooms can be predicted well using numerical procedures such as the method of images and ray tracing. However, this requires acoustical expertise, as well as computational resources and times which result in prediction methods only being used in special cases. This paper discusses alternative empirical prediction methods which have the potential to be sufficiently accurate in ‘‘typical’’ cases, and more readily accessible to practitioners, making them more likely to be used in practice. The first method discussed is a hybrid approach, whereby characteristic workroom sound-propagation curves are predicted using ray tracing. These are then input into an empirical model which sums the energy contribution of all sources at a receiver position based on those curves and the applicable source/receiver distances. Next, the development of empirical models for predicting frequency-varying sound-propagation curves and reverberation times using regression techniques is discussed. These were developed from data measured in actual workrooms when empty or fitted, without and with sound-absorptive treatment. Empirical methods for estimating workroom fitting densities and multisource noise levels, and the integration of the empirical models into the PlantNoise prediction system, are also discussed.