Ray-tracing evaluation of empirical models for predicting noise in industrial workshops

To evaluate the accuracy of empirical models for predicting steady-state noise levels and reverberation times in typical industrial workshops, predictions by these models were compared with predictions by a ray-tracing model, nominally using the same input data. Comparisons were made for three workshops—‘long’, ‘flat’ and ‘quasi-cubic’ in shape—with reflective and absorbent ceilings, when empty and with four densities of fittings. In the case of the ‘long’ room, noise levels were predicted along both the long and short major horizontal axes. In ‘long’ and ‘flat’ workshops, steady-state prediction by the empirical models often agreed with ray-tracing prediction within 2 dB. This result suggests that the empirical models are fundamentally valid. However, agreement was worse at large source/receiver distances, and at 500 Hz. Empirical reverberation-time prediction generally agreed less well with ray tracing, possibly indicating that the empirical reverberation-time prediction models are less valid. Strong disagreement occurred between the models in the case of steady-state prediction in the ‘quasi-cubic’ workshop, indicating that the empirical steady-state models are invalid in this case. Agreement for reverberation time was good with a non-absorbent ceiling, but poor with an absorbent ceiling.